diff options
104 files changed, 933 insertions, 799 deletions
diff --git a/default.nix b/default.nix index 1944de0..59357b2 100644 --- a/default.nix +++ b/default.nix @@ -5,10 +5,11 @@ let bionix = nixpkgs.lib.makeExtensible (self: let callBionix = file: attrs: import file ({ bionix = self; } // attrs); - in with self; { - callBionix = callBionix; + in + with self; { + inherit callBionix; id = x: x; - exec = f: x: y: f x y; + exec = id; exec' = f: exec (_: f) { }; exec'' = f: exec' (_: f) { }; callBionixE = p: exec (callBionix p); @@ -52,15 +53,17 @@ let slurm-run = callPackage ./lib/slurm.nix { }; slurm-exec = f: x: y: - slurm-run x (f (builtins.removeAttrs x [ - "ppn" - "mem" - "walltime" - "partition" - "slurmFlags" - "salloc" - "srun" - ]) y); + slurm-run x (f + (builtins.removeAttrs x [ + "ppn" + "mem" + "walltime" + "partition" + "slurmFlags" + "salloc" + "srun" + ]) + y); slurm = bionix.extend (self: super: { exec = super.slurm-run; }); qsub = attrs: bionix.extend (self: super: @@ -78,7 +81,8 @@ let qsub = attrs: (callPackage ./lib/qsub.nix { }) (qsubDefs // attrs); exec = f: x: y: qsub (builtins.intersectAttrs qsubDefs x) (super.exec f - (builtins.removeAttrs x (builtins.attrNames qsubDefs)) y); + (builtins.removeAttrs x (builtins.attrNames qsubDefs)) + y); }); def = f: defs: attrs: f (defs // attrs); @@ -88,22 +92,24 @@ let name = "link-outputs"; outputs = [ "out" ] ++ attrNames x; nativeBuildInputs = [ pkgs.perl ]; - buildCommand = let - recurse = x: - if x ? type && x.type == "derivation" then - x - else if builtins.typeOf x == "set" then - linkOutputs x - else - abort "linkOutputs: unsupported type"; - link = dst: src: '' - ln -s ${recurse src} $(perl -e 'print $ENV{"${dst}"}') ; ln -s ${ - recurse src - } $out/${dst} - ''; - in '' - mkdir $out - '' + (concatStringsSep "\n" (mapAttrsToList link x)); + buildCommand = + let + recurse = x: + if x ? type && x.type == "derivation" then + x + else if builtins.typeOf x == "set" then + linkOutputs x + else + abort "linkOutputs: unsupported type"; + link = dst: src: '' + ln -s ${recurse src} $(perl -e 'print $ENV{"${dst}"}') ; ln -s ${ + recurse src + } $out/${dst} + ''; + in + '' + mkdir $out + '' + (concatStringsSep "\n" (mapAttrsToList link x)); passthru.linkInputs = x; }; @@ -127,15 +133,15 @@ let # Export nixpkgs and standard library lib pkgs = nixpkgs; lib = nixpkgs.lib // { - types = types; + inherit types; shard = callBionix ./lib/shard.nix { }; }; stage = x@{ name, stripStorePaths ? true, multicore ? false, ... }: (if stripStorePaths then strip else x: x) - (nixpkgs.stdenvNoCC.mkDerivation (x // { - name = "bionix-" + name; - inherit multicore; - })); + (nixpkgs.stdenvNoCC.mkDerivation (x // { + name = "bionix-" + name; + inherit multicore; + })); strip = drv: let stripCommand = '' @@ -156,7 +162,8 @@ let rewriteOutput $o done ''; - in drv.overrideAttrs (attrs: + in + drv.overrideAttrs (attrs: if attrs ? buildCommand then { buildCommand = attrs.buildCommand + stripCommand; } else { @@ -186,13 +193,15 @@ let overlayByType = { lambda = bionix: overlay: bionix.extend - (self: super: nixpkgs.lib.recursiveUpdate super (overlay self super)); + (self: super: nixpkgs.lib.recursiveUpdate super (overlay self super)); path = bionix: path: overlay bionix (import path); }; overlay = bionix: overlay: let overlayType = builtins.typeOf overlay; - in if overlayByType ? ${overlayType} then - overlayByType.${overlayType} bionix overlay + in + if overlayByType ? "${overlayType}" then + overlayByType."${overlayType}" bionix overlay else builtins.throw ("cannot overlay type " + overlayType); -in with nixpkgs.lib; foldl overlay bionix overlays +in +with nixpkgs.lib; foldl overlay bionix overlays diff --git a/doc/default.nix b/doc/default.nix index d68a6b1..ac8bfeb 100644 --- a/doc/default.nix +++ b/doc/default.nix @@ -1,4 +1,4 @@ -{ bionix ? import ./.. {} }: +{ bionix ? import ./.. { } }: with bionix; diff --git a/examples/call.nix b/examples/call.nix index 8a38fdb..81a112e 100644 --- a/examples/call.nix +++ b/examples/call.nix @@ -2,9 +2,10 @@ # with the Platypus variant caller. Each input is preprocessed by aligning # against a reference genome (defaults to GRCH38), fixing mate information, and # marking duplicates. Finally platypus is called over all samples. -{bionix ? import <bionix> {} -,inputs -,ref ? bionix.ref.grch38.seq }: +{ bionix ? import <bionix> { } +, inputs +, ref ? bionix.ref.grch38.seq +}: with bionix; with lib; @@ -13,9 +14,10 @@ let preprocess = flip pipe [ (bwa.align { inherit ref; }) (samtools.sort { nameSort = true; }) - (samtools.fixmate {}) - (samtools.sort {}) - (samtools.markdup {}) + (samtools.fixmate { }) + (samtools.sort { }) + (samtools.markdup { }) ]; -in platypus.call {} (map preprocess inputs) +in +platypus.call { } (map preprocess inputs) diff --git a/examples/default.nix b/examples/default.nix index 9aace38..ef57341 100644 --- a/examples/default.nix +++ b/examples/default.nix @@ -1,6 +1,6 @@ # This example uses the pipelines specified in the call.nix file on the # synthetic data in this directory. -{bionix ? import <bionix> {}}: +{ bionix ? import <bionix> { } }: with bionix; @@ -15,7 +15,7 @@ let sha256 = "0kh29i6fg14dn0fb1xj6pkpk6d83y7zg7aphkbvjrhm82braqkm8"; }; - input2 = fetchFastQ { + input2 = fetchFastQ { url = "https://github.com/PapenfussLab/bionix/raw/master/examples/sample1-2.fq"; sha256 = "0czk85km6a91y0fn4b7f9q7ps19b5jf7jzwbly4sgznps7ir2kdk"; }; @@ -28,7 +28,7 @@ let sha256 = "08gixavfklqvk1m2ic6v56z82vl00qnpsd9xb64z6zl03nz98mcy"; }; - input2 = fetchFastQ { + input2 = fetchFastQ { url = "https://github.com/PapenfussLab/bionix/raw/master/examples/sample2-2.fq"; sha256 = "1xxwm2vq52axpdhm14rh5mg5nzzpxaqnvhzrqhajm27fqksgzjjw"; }; @@ -41,4 +41,5 @@ let sha256 = "0sy9hq8n55knfkiblam50dzaiwhrx6pv8b8l1njdn6kfj4wflz2p"; }; -in import ./call.nix {inherit inputs ref bionix;} +in +import ./call.nix { inherit inputs ref bionix; } diff --git a/examples/ex-nextflow/nextflow-example1.nix b/examples/ex-nextflow/nextflow-example1.nix index 4f00431..687505c 100644 --- a/examples/ex-nextflow/nextflow-example1.nix +++ b/examples/ex-nextflow/nextflow-example1.nix @@ -1,7 +1,8 @@ # This is a translation of the Nextflow example found at # https://www.nextflow.io/example1.html -{ bionix ? import ./../.. {} -, input ? ./sample.fa}: +{ bionix ? import ./../.. { } +, input ? ./sample.fa +}: with bionix; with lib; @@ -24,7 +25,8 @@ let ''; }; -in pipe input [ +in +pipe input [ splitSequences (each reverse) ] diff --git a/examples/ex-tnpair/cluster.nix b/examples/ex-tnpair/cluster.nix index 8d54995..5d4e1de 100644 --- a/examples/ex-tnpair/cluster.nix +++ b/examples/ex-tnpair/cluster.nix @@ -4,11 +4,12 @@ let bionix = import <bionix> { overlays = [ (_: super: - super."${if tmpDir == null then "slurm" else "qsub"}" { - ppn = 24; - mem = 7; - walltime = "3:00:00"; - } // super.lib.optionalAttrs (tmpDir != null) { inherit tmpDir; }) + super."${if tmpDir == null then "slurm" else "qsub"}" + { + ppn = 24; + mem = 7; + walltime = "3:00:00"; + } // super.lib.optionalAttrs (tmpDir != null) { inherit tmpDir; }) (self: super: with super; { @@ -27,4 +28,5 @@ let }) ]; }; -in import ./. { inherit bionix; } +in +import ./. { inherit bionix; } diff --git a/examples/ex-tnpair/default.nix b/examples/ex-tnpair/default.nix index f9581f8..65a2265 100644 --- a/examples/ex-tnpair/default.nix +++ b/examples/ex-tnpair/default.nix @@ -45,6 +45,7 @@ let fetch = s: mapAttrs (_: fetchFastQGZ) s.inputs; -in import ./tnpair.nix { +in +import ./tnpair.nix { inherit pair fetch bionix; } diff --git a/examples/ex-tnpair/tnpair.nix b/examples/ex-tnpair/tnpair.nix index 9e6befe..414a391 100644 --- a/examples/ex-tnpair/tnpair.nix +++ b/examples/ex-tnpair/tnpair.nix @@ -1,4 +1,4 @@ -{bionix ? import <bionix> {}, pair, fetch}: +{ bionix ? import <bionix> { }, pair, fetch }: with bionix; with lib; @@ -13,7 +13,7 @@ let fetch (align { preset = "sr"; ref = ref.grch38.seq; flags = "-R'@RG\\tID:${s.type}\\tSM:${s.type}'"; }) (sort { nameSort = true; }) - (fixmate {}) + (fixmate { }) (sort { }) (markdup { }) ]; @@ -28,21 +28,26 @@ let bams = mapAttrs (_: preprocess) pair; - variants = let - somatic = strelka.callSomatic { } bams; in mapAttrs (_: flip pipe [ - (compression.uncompress { }) - (snpeff.annotate { db = ref.grch38.snpeff.db; }) - dropErrors - (snpeff.dbnsfp { dbnsfp = ref.grch38.snpeff.dbnsfp; }) - ]) { - "snvs.vcf" = somatic.snvs; - "indels.vcf" = somatic.snvs; - "germline.vcf" = strelka.call { } [bams.normal]; - }; + variants = + let + somatic = strelka.callSomatic { } bams; in + mapAttrs + (_: flip pipe [ + (compression.uncompress { }) + (snpeff.annotate { inherit (ref.grch38.snpeff) db; }) + dropErrors + (snpeff.dbnsfp { inherit (ref.grch38.snpeff) dbnsfp; }) + ]) + { + "snvs.vcf" = somatic.snvs; + "indels.vcf" = somatic.snvs; + "germline.vcf" = strelka.call { } [ bams.normal ]; + }; cnvs = cnvkit.callCNV { } { normals = [ bams.normal ]; tumours = [ bams.tumour ]; }; -in linkOutputs { +in +linkOutputs { inherit variants; alignments = linkOutputs (mapAttrs' (n: nameValuePair (n + ".bam")) bams); cnvkit = cnvs; diff --git a/examples/ex-wdl/wdl-scatter-gather.nix b/examples/ex-wdl/wdl-scatter-gather.nix index 61dc441..5245662 100644 --- a/examples/ex-wdl/wdl-scatter-gather.nix +++ b/examples/ex-wdl/wdl-scatter-gather.nix @@ -1,6 +1,6 @@ # The scatter-gather example from https://github.com/openwdl/wdl # translated to bionix -{ bionix ? import ./../.. {} }: +{ bionix ? import ./../.. { } }: with bionix; with lib; @@ -30,4 +30,5 @@ let ''; }; -in gather (map analysis prepare) +in +gather (map analysis prepare) diff --git a/lib/google.nix b/lib/google.nix index 354b5c5..06b80b3 100644 --- a/lib/google.nix +++ b/lib/google.nix @@ -10,7 +10,9 @@ let "https://raw.githubusercontent.com/circulosmeos/gdown.pl/master/gdown.pl"; sha256 = "1pw3vg70bgf33akbbphpr6zn3jndv0khmsa3k0m877hgzg1v52qv"; }; -in runCommand "gdown" { +in +runCommand "gdown" +{ nativeBuildInputs = [ perl wget ]; outputHashAlgo = "sha256"; outputHash = sha256; diff --git a/lib/qsub.nix b/lib/qsub.nix index a8827a8..2528162 100644 --- a/lib/qsub.nix +++ b/lib/qsub.nix @@ -4,76 +4,87 @@ with lib; let escape = x: if builtins.typeOf x == "string" then escapeShellArg x else x; -in { ppn, mem, walltime, queue ? null, qsubFlags ? null, tmpDir, sleepTime -, qsubPath ? "/usr/bin" }: +in +{ ppn +, mem +, walltime +, queue ? null +, qsubFlags ? null +, tmpDir +, sleepTime +, qsubPath ? "/usr/bin" +}: drv: let ppnReified = if drv.multicore then ppn else 1; -in overrideDerivation drv ({ args, builder, name, ... }: { +in +overrideDerivation drv ({ args, builder, name, ... }: { builder = "/bin/bash"; - args = let - script = writeScript "qsub-script" '' - #!${stdenv.shell} - while [ ! -e ${tmpDir}/qsub-$PBS_JOBID ] ; do - sleep ${toString sleepTime} - done - set -a - . ${tmpDir}/qsub-$PBS_JOBID/nix-set - set +a - TMPDIR=${tmpDir}/qsub-$PBS_JOBID - TEMP=$TMPDIR - TMP=$TMPDIR - NIX_BUILD_TOP=$TMPDIR - cd $TMPDIR - ${builder} ${concatMapStringsSep " " escape args} &> qsub-log - echo $? > qsub-exit - ''; + args = + let + script = writeScript "qsub-script" '' + #!${stdenv.shell} + while [ ! -e ${tmpDir}/qsub-$PBS_JOBID ] ; do + sleep ${toString sleepTime} + done + set -a + . ${tmpDir}/qsub-$PBS_JOBID/nix-set + set +a + TMPDIR=${tmpDir}/qsub-$PBS_JOBID + TEMP=$TMPDIR + TMP=$TMPDIR + NIX_BUILD_TOP=$TMPDIR + cd $TMPDIR + ${builder} ${concatMapStringsSep " " escape args} &> qsub-log + echo $? > qsub-exit + ''; - qsub = writeScript "qsub" '' - #!${stdenv.shell} - PATH=${qsubPath} - SHELL=/bin/sh - NIX_BUILD_CORES=${toString ppnReified} + qsub = writeScript "qsub" '' + #!${stdenv.shell} + PATH=${qsubPath} + SHELL=/bin/sh + NIX_BUILD_CORES=${toString ppnReified} - while : ; do - qsub -l nodes=1:ppn=${toString ppnReified},mem=${ - toString mem - }gb,walltime=${walltime} \ - -N "${name}" \ - ${optionalString (queue != null) "-q ${queue}"} \ - ${optionalString (qsubFlags != null) qsubFlags} \ - ${script} 2>&1 > id - if [ $? -eq 0 ] ; then - break - fi - if ! grep "Please retry" id > /dev/null ; then - cat id >&2 - exit 1 - fi - sleep ${toString sleepTime} - done - id=$(cat id) - echo $id + while : ; do + qsub -l nodes=1:ppn=${toString ppnReified},mem=${ + toString mem + }gb,walltime=${walltime} \ + -N "${name}" \ + ${optionalString (queue != null) "-q ${queue}"} \ + ${optionalString (qsubFlags != null) qsubFlags} \ + ${script} 2>&1 > id + if [ $? -eq 0 ] ; then + break + fi + if ! grep "Please retry" id > /dev/null ; then + cat id >&2 + exit 1 + fi + sleep ${toString sleepTime} + done + id=$(cat id) + echo $id - function cleanup { - qdel $id 2>/dev/null || true - sleep ${toString sleepTime} - rm -rf ${tmpDir}/qsub-$id - } - trap cleanup INT TERM EXIT + function cleanup { + qdel $id 2>/dev/null || true + sleep ${toString sleepTime} + rm -rf ${tmpDir}/qsub-$id + } + trap cleanup INT TERM EXIT - cp -r $TMPDIR ${tmpDir}/qsub-$id - set > ${tmpDir}/qsub-$id/nix-set - until qstat -f ''${id%%.} 2>&1 | grep "\(Unknown Job\|job_state = C\)" > /dev/null ; do - sleep ${toString sleepTime} - done - cat ${tmpDir}/qsub-$id/qsub-log - if [ -e ${tmpDir}/qsub-$id/qsub-exit ]; then - exitCode=$(cat ${tmpDir}/qsub-$id/qsub-exit) - else - exitCode=1 - fi - exit $exitCode - ''; + cp -r $TMPDIR ${tmpDir}/qsub-$id + set > ${tmpDir}/qsub-$id/nix-set + until qstat -f ''${id%%.} 2>&1 | grep "\(Unknown Job\|job_state = C\)" > /dev/null ; do + sleep ${toString sleepTime} + done + cat ${tmpDir}/qsub-$id/qsub-log + if [ -e ${tmpDir}/qsub-$id/qsub-exit ]; then + exitCode=$(cat ${tmpDir}/qsub-$id/qsub-exit) + else + exitCode=1 + fi + exit $exitCode + ''; - in [ "-c" qsub ]; + in + [ "-c" qsub ]; }) diff --git a/lib/references.nix b/lib/references.nix index 23ccde3..fd8c3cc 100644 --- a/lib/references.nix +++ b/lib/references.nix @@ -26,7 +26,7 @@ rec { buildInputs = with pkgs; [ gawk ]; buildCommand = '' gunzip < $src | awk '/^[^#]/{print "chr" $0;next}{print}' > $out - ''; + ''; passthru.filetype = filetype.vcf { ref = seq; }; }; encode.blacklist = stage { @@ -42,28 +42,30 @@ rec { ''; }; - ensembl = let version = "74"; - in { - cdna = stage { - name = "ensembl-grch37-cdna-${version}"; - src = pkgs.fetchurl { - url = - "ftp://ftp.ensembl.org/pub/release-${version}/fasta/homo_sapiens/cdna/Homo_sapiens.GRCh37.${version}.cdna.all.fa.gz"; - sha256 = "1m62hiw17zcxg3sza0aq53885wb8g202j8lc1ilhmkg2izzbyihj"; + ensembl = + let version = "74"; + in + { + cdna = stage { + name = "ensembl-grch37-cdna-${version}"; + src = pkgs.fetchurl { + url = + "ftp://ftp.ensembl.org/pub/release-${version}/fasta/homo_sapiens/cdna/Homo_sapiens.GRCh37.${version}.cdna.all.fa.gz"; + sha256 = "1m62hiw17zcxg3sza0aq53885wb8g202j8lc1ilhmkg2izzbyihj"; + }; + buildCommand = "gunzip < $src > $out"; + passthru.filetype = filetype.fa { }; }; - buildCommand = "gunzip < $src > $out"; - passthru.filetype = filetype.fa { }; - }; - gtf = stage { - name = "ensembl-grch37-gtf-${version}"; - src = pkgs.fetchurl { - url = - "ftp://ftp.ensembl.org/pub/release-${version}/gtf/homo_sapiens/Homo_sapiens.GRCh37.${version}.gtf.gz"; - sha256 = "1m62hiw17zcxg3sza0aq53885wb8g202j8lc1ilhmkg2izzbyihj"; + gtf = stage { + name = "ensembl-grch37-gtf-${version}"; + src = pkgs.fetchurl { + url = + "ftp://ftp.ensembl.org/pub/release-${version}/gtf/homo_sapiens/Homo_sapiens.GRCh37.${version}.gtf.gz"; + sha256 = "1m62hiw17zcxg3sza0aq53885wb8g202j8lc1ilhmkg2izzbyihj"; + }; + buildCommand = "gunzip < $src > $out"; }; - buildCommand = "gunzip < $src > $out"; }; - }; snpeff = { db = pkgs.stdenv.mkDerivation rec { name = "GRCh37.75"; @@ -114,7 +116,7 @@ rec { buildInputs = with pkgs; [ gawk ]; buildCommand = '' gunzip < $src | awk '/^[^#]/{print "chr" $0;next}{print}' > $out - ''; + ''; passthru.filetype = filetype.vcf { ref = seq; }; }; encode.blacklist = stage { @@ -129,47 +131,49 @@ rec { gunzip < $bed | awk '{print $1 ":" $2 "-" $3}' > $out ''; }; - ensembl = let version = "97"; - in { - cdna = stage { - name = "ensembl-grch38-cdna-${version}"; - src = pkgs.fetchurl { - url = - "ftp://ftp.ensembl.org/pub/release-${version}/fasta/homo_sapiens/cdna/Homo_sapiens.GRCh38.cdna.all.fa.gz"; - sha256 = "1m6hvnvlrsi6bzcmq0lnv0igy3in1a7jp723yc74g4g6zjp3cy8c"; + ensembl = + let version = "97"; + in + { + cdna = stage { + name = "ensembl-grch38-cdna-${version}"; + src = pkgs.fetchurl { + url = + "ftp://ftp.ensembl.org/pub/release-${version}/fasta/homo_sapiens/cdna/Homo_sapiens.GRCh38.cdna.all.fa.gz"; + sha256 = "1m6hvnvlrsi6bzcmq0lnv0igy3in1a7jp723yc74g4g6zjp3cy8c"; + }; + buildCommand = "gunzip < $src > $out"; + passthru.filetype = filetype.fa { }; }; - buildCommand = "gunzip < $src > $out"; - passthru.filetype = filetype.fa { }; - }; - ncrna = stage { - name = "ensembl-grch38-ncrna-${version}"; - src = pkgs.fetchurl { - url = - "ftp://ftp.ensembl.org/pub/release-${version}/fasta/homo_sapiens/ncrna/Homo_sapiens.GRCh38.ncrna.fa.gz"; - sha256 = "1r0dmybn31wf6xc90z2c08ngivkv39hqa8wqg3vik6s4spwpdhj0"; + ncrna = stage { + name = "ensembl-grch38-ncrna-${version}"; + src = pkgs.fetchurl { + url = + "ftp://ftp.ensembl.org/pub/release-${version}/fasta/homo_sapiens/ncrna/Homo_sapiens.GRCh38.ncrna.fa.gz"; + sha256 = "1r0dmybn31wf6xc90z2c08ngivkv39hqa8wqg3vik6s4spwpdhj0"; + }; + buildCommand = "gunzip < $src > $out"; + passthru.filetype = filetype.fa { }; }; - buildCommand = "gunzip < $src > $out"; - passthru.filetype = filetype.fa { }; - }; - gtf = stage { - name = "ensembl-grch38-gtf-${version}"; - src = pkgs.fetchurl { - url = - "ftp://ftp.ensembl.org/pub/release-${version}/gtf/homo_sapiens/Homo_sapiens.GRCh38.${version}.gtf.gz"; - sha256 = "068ab5jf87il301jcr0576c4q0smv6kxpv94gnrm3qzl6kvmaawd"; + gtf = stage { + name = "ensembl-grch38-gtf-${version}"; + src = pkgs.fetchurl { + url = + "ftp://ftp.ensembl.org/pub/release-${version}/gtf/homo_sapiens/Homo_sapiens.GRCh38.${version}.gtf.gz"; + sha256 = "068ab5jf87il301jcr0576c4q0smv6kxpv94gnrm3qzl6kvmaawd"; + }; + buildCommand = "gunzip < $src > $out"; }; - buildCommand = "gunzip < $src > $out"; - }; - gff3 = stage { - name = "ensembl-grch38-gff-${version}"; - src = pkgs.fetchurl { - url = - "ftp://ftp.ensembl.org/pub/release-${version}/gff3/homo_sapiens/Homo_sapiens.GRCh38.${version}.gff3.gz"; - sha256 = "1xvhsn938mw0032qgc9dvw3k2xrhpx77b8ms03fkrs2s67f7zli7"; + gff3 = stage { + name = "ensembl-grch38-gff-${version}"; + src = pkgs.fetchurl { + url = + "ftp://ftp.ensembl.org/pub/release-${version}/gff3/homo_sapiens/Homo_sapiens.GRCh38.${version}.gff3.gz"; + sha256 = "1xvhsn938mw0032qgc9dvw3k2xrhpx77b8ms03fkrs2s67f7zli7"; + }; + buildCommand = "gunzip < $src > $out"; }; - buildCommand = "gunzip < $src > $out"; }; - }; snpeff = { db = pkgs.stdenv.mkDerivation rec { name = "GRCh38.86"; @@ -221,48 +225,50 @@ rec { buildCommand = "gunzip < $src > $out"; passthru.filetype = filetype.fa { }; }; - ensembl = let version = "94"; - in { - cdna = stage { - name = "ensembl-grcm38-cdna-${version}"; - src = pkgs.fetchurl { - url = - "ftp://ftp.ensembl.org/pub/release-${version}/fasta/mus_musculus/cdna/Mus_musculus.GRCm38.cdna.all.fa.gz"; - sha256 = "0khp9l6s35lav2xqp7vkk6ybnz4wjihn7lapjf2lbpnbzjb4hp6d"; + ensembl = + let version = "94"; + in + { + cdna = stage { + name = "ensembl-grcm38-cdna-${version}"; + src = pkgs.fetchurl { + url = + "ftp://ftp.ensembl.org/pub/release-${version}/fasta/mus_musculus/cdna/Mus_musculus.GRCm38.cdna.all.fa.gz"; + sha256 = "0khp9l6s35lav2xqp7vkk6ybnz4wjihn7lapjf2lbpnbzjb4hp6d"; + }; + buildCommand = "gunzip < $src > $out"; + passthru.filetype = filetype.fa { }; }; - buildCommand = "gunzip < $src > $out"; - passthru.filetype = filetype.fa { }; - }; - ncrna = stage { - name = "ensembl-grcm38-ncrna-${version}"; - version = "94"; - src = pkgs.fetchurl { - url = - "ftp://ftp.ensembl.org/pub/release-${version}/fasta/mus_musculus/ncrna/Mus_musculus.GRCm38.ncrna.fa.gz"; - sha256 = "0d997gm8p2b89rm5d46m2x4vz9lijxarfr2lzylnbi8gyqrbagdd"; + ncrna = stage { + name = "ensembl-grcm38-ncrna-${version}"; + version = "94"; + src = pkgs.fetchurl { + url = + "ftp://ftp.ensembl.org/pub/release-${version}/fasta/mus_musculus/ncrna/Mus_musculus.GRCm38.ncrna.fa.gz"; + sha256 = "0d997gm8p2b89rm5d46m2x4vz9lijxarfr2lzylnbi8gyqrbagdd"; + }; + buildCommand = "gunzip < $src > $out"; + passthru.filetype = filetype.fa { }; }; - buildCommand = "gunzip < $src > $out"; - passthru.filetype = filetype.fa { }; - }; - gtf = stage { - name = "ensembl-grcm38-gtf-${version}"; - src = pkgs.fetchurl { - url = - "ftp://ftp.ensembl.org/pub/release-${version}/gtf/mus_musculus/Mus_musculus.GRCm38.${version}.gtf.gz"; - sha256 = "0i61jq5i5bcini5nxqxxp3rnz2xzgychvzdn0k451f5rv053lp3v"; + gtf = stage { + name = "ensembl-grcm38-gtf-${version}"; + src = pkgs.fetchurl { + url = + "ftp://ftp.ensembl.org/pub/release-${version}/gtf/mus_musculus/Mus_musculus.GRCm38.${version}.gtf.gz"; + sha256 = "0i61jq5i5bcini5nxqxxp3rnz2xzgychvzdn0k451f5rv053lp3v"; + }; + buildCommand = "gunzip < $src > $out"; }; - buildCommand = "gunzip < $src > $out"; - }; - gff3 = stage { - name = "ensembl-grcm38-gff-${version}"; - src = pkgs.fetchurl { - url = - "ftp://ftp.ensembl.org/pub/release-${version}/gff3/mus_musculus/Mus_musculus.GRCm38.${version}.gff3.gz"; - sha256 = "15fmdpx6g96fygwhs10jwrb2q5p9y64bc3d4clg856k57qzzgprg"; + gff3 = stage { + name = "ensembl-grcm38-gff-${version}"; + src = pkgs.fetchurl { + url = + "ftp://ftp.ensembl.org/pub/release-${version}/gff3/mus_musculus/Mus_musculus.GRCm38.${version}.gff3.gz"; + sha256 = "15fmdpx6g96fygwhs10jwrb2q5p9y64bc3d4clg856k57qzzgprg"; + }; + buildCommand = "gunzip < $src > $out"; }; - buildCommand = "gunzip < $src > $out"; }; - }; }; mm10 = mm10-p4; diff --git a/lib/shard-regex.nix b/lib/shard-regex.nix index 06fba4d..9dc1720 100644 --- a/lib/shard-regex.nix +++ b/lib/shard-regex.nix @@ -1,54 +1,64 @@ -{bionix, n}: +{ bionix, n }: with bionix; with lib.types; input: let - re = let f = matchFiletype' "shard-regex" { + re = + let f = matchFiletype' "shard-regex" { fa = _: "^>"; fq = _: "^@"; gz = f; bz2 = f; }; - in f input.filetype; - decompress = matchFiletype "shard-regex-decompression" { + in f input.filetype; + decompress = matchFiletype "shard-regex-decompression" + { fa = _: "cat"; fq = _: "cat"; gz = _: "gunzip"; bz2 = _: "bunzip2"; - } input; - compress = matchFiletype "shard-regex-compression" { + } + input; + compress = matchFiletype "shard-regex-compression" + { fa = _: "cat"; fq = _: "cat"; gz = _: "gzip"; bz2 = _: "bzip2"; - } input; - compressPkgs = with bionix.pkgs; matchFiletype "shard-regex-compression" { - fa = _: []; - fq = _: []; + } + input; + compressPkgs = with bionix.pkgs; matchFiletype "shard-regex-compression" + { + fa = _: [ ]; + fq = _: [ ]; gz = _: [ gzip ]; bz2 = _: [ bzip2 ]; - } input; -in stage { + } + input; +in +stage { name = "shard"; outputs = [ "out" ] ++ builtins.genList (i: "out" + toString (i + 2)) (n - 1); buildInputs = [ pkgs.gawk ] ++ compressPkgs; - buildCommand = let - awkScript = pkgs.writeText "shard.awk" '' - BEGIN{cout=0} - FNR==NR{out[nout++] = $0;next} - /${re}/{cout = (cout + 1) % nout} - {print > out[cout]} + buildCommand = + let + awkScript = pkgs.writeText "shard.awk" '' + BEGIN{cout=0} + FNR==NR{out[nout++] = $0;next} + /${re}/{cout = (cout + 1) % nout} + {print > out[cout]} + ''; + in + '' + for o in $outputs ; do + echo $(basename ''${!o}) >> outputs + done + awk -f ${awkScript} outputs <(${decompress} < ${input}) + for o in $outputs ; do + ${compress} < $(basename ''${!o}) > ''${!o} + done ''; - in '' - for o in $outputs ; do - echo $(basename ''${!o}) >> outputs - done - awk -f ${awkScript} outputs <(${decompress} < ${input}) - for o in $outputs ; do - ${compress} < $(basename ''${!o}) > ''${!o} - done - ''; passthru.filetype = input.filetype; } diff --git a/lib/shard.nix b/lib/shard.nix index 95cf251..9b85027 100644 --- a/lib/shard.nix +++ b/lib/shard.nix @@ -1,9 +1,9 @@ -{bionix}: +{ bionix }: with bionix; with lib.types; { - regex = n: input: outputDrvs (callBionixE ./shard-regex.nix { n = n; } input); - fastQPair = n: {input1, input2}: lib.zipListsWith (i: j: {input1 = i; input2 = j;}) (lib.shard.regex n input1) (lib.shard.regex n input2); + regex = n: input: outputDrvs (callBionixE ./shard-regex.nix { inherit n; } input); + fastQPair = n: { input1, input2 }: lib.zipListsWith (i: j: { input1 = i; input2 = j; }) (lib.shard.regex n input1) (lib.shard.regex n input2); } diff --git a/lib/slurm.nix b/lib/slurm.nix index c050628..8c9ad50 100644 --- a/lib/slurm.nix +++ b/lib/slurm.nix @@ -4,23 +4,34 @@ with lib; let escape = x: if builtins.typeOf x == "string" then escapeShellArg x else x; -in { ppn, mem, walltime, partition ? null, slurmFlags ? null -, salloc ? "/usr/bin/salloc", srun ? "/usr/bin/srun", ... }: +in +{ ppn +, mem +, walltime +, partition ? null +, slurmFlags ? null +, salloc ? "/usr/bin/salloc" +, srun ? "/usr/bin/srun" +, ... +}: drv: let ppnReified = if drv.multicore then ppn else 1; -in overrideDerivation drv ({ args, builder, name, ... }: { +in +overrideDerivation drv ({ args, builder, name, ... }: { builder = stdenv.shell; - args = let - slurm = writeScript "slurm" '' - #!${stdenv.shell} - NIX_BUILD_CORES=${toString ppnReified} + args = + let + slurm = writeScript "slurm" '' + #!${stdenv.shell} + NIX_BUILD_CORES=${toString ppnReified} - ${salloc} -c $NIX_BUILD_CORES --mem=${toString mem}G -t ${walltime} \ - -J "${name}" \ - ${optionalString (partition != null) "-p ${partition}"} \ - ${optionalString (slurmFlags != null) slurmFlags} \ - ${srun} ${builder} ${concatMapStringsSep " " escape args} - ''; + ${salloc} -c $NIX_BUILD_CORES --mem=${toString mem}G -t ${walltime} \ + -J "${name}" \ + ${optionalString (partition != null) "-p ${partition}"} \ + ${optionalString (slurmFlags != null) slurmFlags} \ + ${srun} ${builder} ${concatMapStringsSep " " escape args} + ''; - in [ "-c" slurm ]; + in + [ "-c" slurm ]; }) diff --git a/lib/types.nix b/lib/types.nix index fbc7513..7e58ce0 100644 --- a/lib/types.nix +++ b/lib/types.nix @@ -1,4 +1,4 @@ -{bionix}: +{ bionix }: with bionix; with lib; @@ -25,30 +25,30 @@ rec { matchFiletype = sym: y: x: if x ? filetype then matchFiletype' sym y x.filetype else abort "unknown filetype for ${sym}"; matchFiletype' = sym: y: x: match x (defError (idft sym) y filetype); filetype = make-type "filetype" { - fa = {}; - fq = {}; - bam = {ref = any; sorting = sort;}; - sam = {ref = any; sorting = sort;}; - cram = {ref = any; sorting = sort;}; - vcf = {ref = any;}; - bed = {ref = any;}; + fa = { }; + fq = { }; + bam = { ref = any; sorting = sort; }; + sam = { ref = any; sorting = sort; }; + cram = { ref = any; sorting = sort; }; + vcf = { ref = any; }; + bed = { ref = any; }; gz = filetype; bz2 = filetype; }; - toCram = matchFiletype "bam2cram" { bam = filetype.cram; sam = filetype.cram; cram = filetype.cram; }; - toBam = matchFiletype "bam2cram" { bam = filetype.bam; sam = filetype.bam; cram = filetype.bam; }; - toSam = matchFiletype "bam2cram" { bam = filetype.sam; sam = filetype.sam; cram = filetype.sam; }; + toCram = matchFiletype "bam2cram" { bam = filetype.cram; sam = filetype.cram; inherit (filetype) cram; }; + toBam = matchFiletype "bam2cram" { inherit (filetype) bam; sam = filetype.bam; cram = filetype.bam; }; + toSam = matchFiletype "bam2cram" { bam = filetype.sam; inherit (filetype) sam; cram = filetype.sam; }; matchSorting = sym: y: x: match x.sorting (defError (idst sym) y sort); matchFileSorting = sym: y: let f = matchSorting sym y; in matchFiletype sym { bam = f; sam = f; cram = f; }; sort = make-type "sort" { - none = {}; - coord = {}; - name = {}; + none = { }; + coord = { }; + name = { }; }; - coordSort = f: matchFiletype "coordSort" { bam = x: filetype.bam (x // {sorting = sort.coord {};}); } {filetype = f;}; - nameSort = f: matchFiletype "nameSort" { bam = x: filetype.bam (x // {sorting = sort.name {};}); } {filetype = f;}; + coordSort = f: matchFiletype "coordSort" { bam = x: filetype.bam (x // { sorting = sort.coord { }; }); } { filetype = f; }; + nameSort = f: matchFiletype "nameSort" { bam = x: filetype.bam (x // { sorting = sort.name { }; }); } { filetype = f; }; gunzip = matchFiletype "gunzip" { gz = x: x; }; bunzip2 = matchFiletype "bunzip2" { bz2 = x: x; }; diff --git a/test-tnpair.nix b/test-tnpair.nix index 6eb2e23..9ba0e51 100644 --- a/test-tnpair.nix +++ b/test-tnpair.nix @@ -1,4 +1,4 @@ -with import <bionix> {}; +with import <bionix> { }; with lib; let @@ -6,28 +6,32 @@ let name = baseNameOf path; buildCommand = "ln -s ${path} $out"; }; - fetchfq = attrs: types.tagFiletype (types.filetype.fq {}) (fetchlocal attrs); - fetchfa = attrs: types.tagFiletype (types.filetype.fa {}) (fetchlocal attrs); + fetchfq = attrs: types.tagFiletype (types.filetype.fq { }) (fetchlocal attrs); + fetchfa = attrs: types.tagFiletype (types.filetype.fa { }) (fetchlocal attrs); ref = fetchfa ./examples/ref.fa; - alignWithRG = rg: x: nameSort (bwa.align { inherit ref; flags = "-R'@RG\\tID:${rg}\\tSM:${rg}'";} x); - sort = samtools.sort {}; - nameSort = samtools.sort {nameSort = true;}; - flagstat = samtools.flagstat {}; - check-fastqc = fastqc.check {}; - check-fastp = fastp.check {}; - callVariants = strelka.callSomatic {}; - markdup = samtools.markdup {}; - fixmate = samtools.fixmate {}; + alignWithRG = rg: x: nameSort (bwa.align { inherit ref; flags = "-R'@RG\\tID:${rg}\\tSM:${rg}'"; } x); + sort = sambamba.sort { }; + nameSort = sambamba.sort { nameSort = true; }; + flagstat = samtools.flagstat { }; + check-fastqc = fastqc.check { }; + check-fastp = fastp.check { }; + callVariants = strelka.callSomatic { }; + markdup = samtools.markdup { }; + fixmate = samtools.fixmate { }; tnpair = { - tumour = {name = "mysample1"; files = { + tumour = { + name = "mysample1"; + files = { input1 = fetchfq ./examples/sample1-1.fq; input2 = fetchfq ./examples/sample1-2.fq; }; }; - normal = {name = "mysample2"; files = { + normal = { + name = "mysample2"; + files = { input1 = fetchfq ./examples/sample2-1.fq; input2 = fetchfq ./examples/sample2-2.fq; }; @@ -37,36 +41,36 @@ let processPair = { tumour, normal }: rec { alignments = mapAttrs (_: x: markdup (sort (fixmate (alignWithRG x.name x.files)))) { inherit normal tumour; }; variants = callVariants alignments; - octopusSomatic = octopus.callSomatic {} {inherit (alignments) normal; tumours = [ alignments.tumour ];}; - glvariants = strelka.call {} (builtins.attrValues alignments); - platypusVars = platypus.call {} (builtins.attrValues alignments); - octopusVars = octopus.call {} (builtins.attrValues alignments); - shards = map (x: nameSort (bwa.align {inherit ref;} x)) (shard.fastQPair 2 normal.files); + octopusSomatic = octopus.callSomatic { } { inherit (alignments) normal; tumours = [ alignments.tumour ]; }; + glvariants = strelka.call { } (builtins.attrValues alignments); + platypusVars = platypus.call { } (builtins.attrValues alignments); + octopusVars = octopus.call { } (builtins.attrValues alignments); + shards = map (x: nameSort (bwa.align { inherit ref; } x)) (shard.fastQPair 2 normal.files); }; tnpairResult = processPair tnpair; cnvkitResults = rec { - cnvs = cnvkit.callCNV {} (with tnpairResult.alignments; { normals = [ normal ]; tumours = [ tumour ];}); - plot = cnvkit.scatterPlot {} cnvs.out1; + cnvs = cnvkit.callCNV { } (with tnpairResult.alignments; { normals = [ normal ]; tumours = [ tumour ]; }); + plot = cnvkit.scatterPlot { } cnvs.out1; }; alignments = { - "bowtie-normal.bam" = bowtie.align {inherit ref;} tnpair.normal.files; - "last.maf" = lastal.align {inherit ref;} tnpair.normal.files; - "whisper.bam" = whisper.align {inherit ref;} tnpair.normal.files; - "bwa-mem.bam" = bwa.mem {inherit ref;} tnpair.normal.files; - "bwa-mem2.bam" = bwa.mem2 {inherit ref;} tnpair.normal.files; - "minimap2-normal.bam" = minimap2.align {inherit ref; preset = "sr"; } tnpair.normal.files; - "snap-normal.bam" = snap.align {inherit ref; } tnpair.normal.files; + "bowtie-normal.bam" = bowtie.align { inherit ref; } tnpair.normal.files; + "last.maf" = lastal.align { inherit ref; } tnpair.normal.files; + "whisper.bam" = whisper.align { inherit ref; } tnpair.normal.files; + "bwa-mem.bam" = bwa.mem { inherit ref; } tnpair.normal.files; + "bwa-mem2.bam" = bwa.mem2 { inherit ref; } tnpair.normal.files; + "minimap2-normal.bam" = minimap2.align { inherit ref; preset = "sr"; } tnpair.normal.files; + "snap-normal.bam" = snap.align { inherit ref; } tnpair.normal.files; "${tnpair.tumour.name}.flagstat" = flagstat tnpairResult.alignments.tumour; - "hisat2-normal.bam" = hisat2.align {inherit ref;} tnpair.normal.files; + "hisat2-normal.bam" = hisat2.align { inherit ref; } tnpair.normal.files; }; testNaming = linkOutputs { - crai = samtools.index {} (samtools.view { outfmt = types.toCram; } tnpairResult.alignments.tumour); - kallisto = kallisto.quant {inherit ref;} (attrValues tnpair.tumour.files); - facets = facets.callCNV {} {vcf = tnpairResult.platypusVars; bams = with tnpairResult.alignments; [ normal tumour ];}; + crai = samtools.index { } (samtools.view { outfmt = types.toCram; } tnpairResult.alignments.tumour); + kallisto = kallisto.quant { inherit ref; } (attrValues tnpair.tumour.files); + facets = facets.callCNV { } { vcf = tnpairResult.platypusVars; bams = with tnpairResult.alignments; [ normal tumour ]; }; cnvkit = cnvkitResults.cnvs; "cnvkit.pdf" = cnvkitResults.plot; "octopus.vcf" = tnpairResult.octopusVars; @@ -74,24 +78,25 @@ let strelka-indels = tnpairResult.variants.indels; "strelka.snvs.vcf" = tnpairResult.variants.snvs; "strelka.gl.vcf" = tnpairResult.glvariants; - delly = delly.call {} (with tnpairResult.alignments; [normal tumour]); - lumpy = lumpy.call {} (with tnpairResult.alignments; [normal tumour]); - manta = manta.call {} (with tnpairResult.alignments; {normals = [normal tumour]; }); - mantaTN = manta.call {} (with tnpairResult.alignments; {normals = [normal]; tumour = tumour;}); - mantaT = manta.call {} (with tnpairResult.alignments; {tumour = tumour;}); - gridss = gridss.callVariants {} (with tnpairResult.alignments; [normal tumour]); - gridss2 = gridss.call (with tnpairResult.alignments; [normal tumour]); - gridss3 = gridss.callAndAssemble (with tnpairResult.alignments; [normal tumour]); - "merged-shards.bam" = samtools.merge {} tnpairResult.shards; - "merged.bam" = samtools.merge {} [tnpairResult.alignments.tumour tnpairResult.alignments.normal]; - "merged-namesorted.bam" = samtools.merge {} [(nameSort tnpairResult.alignments.tumour) (nameSort tnpairResult.alignments.normal)]; - "${tnpair.tumour.name}.cram" = samtools.view { outfmt = types.toCram; } (tnpairResult.alignments.tumour); + delly = delly.call { } (with tnpairResult.alignments; [ normal tumour ]); + lumpy = lumpy.call { } (with tnpairResult.alignments; [ normal tumour ]); + manta = manta.call { } (with tnpairResult.alignments; { normals = [ normal tumour ]; }); + mantaTN = manta.call { } (with tnpairResult.alignments; { normals = [ normal ]; inherit tumour; }); + mantaT = manta.call { } (with tnpairResult.alignments; { inherit tumour; }); + gridss = gridss.callVariants { } (with tnpairResult.alignments; [ normal tumour ]); + gridss2 = gridss.call (with tnpairResult.alignments; [ normal tumour ]); + gridss3 = gridss.callAndAssemble (with tnpairResult.alignments; [ normal tumour ]); + "merged-shards.bam" = samtools.merge { } tnpairResult.shards; + "merged.bam" = samtools.merge { } [ tnpairResult.alignments.tumour tnpairResult.alignments.normal ]; + "merged-namesorted.bam" = samtools.merge { } [ (nameSort tnpairResult.alignments.tumour) (nameSort tnpairResult.alignments.normal) ]; + "${tnpair.tumour.name}.cram" = samtools.view { outfmt = types.toCram; } tnpairResult.alignments.tumour; "${tnpair.tumour.name}.fastqc.1" = check-fastqc tnpair.tumour.files.input1; "${tnpair.tumour.name}.fastp" = check-fastp tnpair.tumour.files; - snver = snver.call { ploidy=1; } tnpairResult.alignments; - mosdepth = mosdepth.plot {} { inputs = mapAttrsToList (_: mosdepth.depth {}) tnpairResult.alignments; names = [ "seq1" "seq2" ]; }; - xenomapper = xenomapper.allocate {} { primary = nameSort tnpairResult.alignments.tumour; secondary = nameSort tnpairResult.alignments.tumour; }; + snver = snver.call { ploidy = 1; } tnpairResult.alignments; + mosdepth = mosdepth.plot { } { inputs = mapAttrsToList (_: mosdepth.depth { }) tnpairResult.alignments; names = [ "seq1" "seq2" ]; }; + xenomapper = xenomapper.allocate { } { primary = nameSort tnpairResult.alignments.tumour; secondary = nameSort tnpairResult.alignments.tumour; }; inherit alignments; }; -in testNaming +in +testNaming diff --git a/tools/ascat-app.nix b/tools/ascat-app.nix index 51aa677..bd6d538 100644 --- a/tools/ascat-app.nix +++ b/tools/ascat-app.nix @@ -1,4 +1,4 @@ -{stdenv, callPackage, buildPerlPackage, fetchurl, fetchFromGitHub, perlPackages, R, bwa, samtools, pkgconfig, zlib, htslib, curl, bzip2, lzma, gnutls, nettle, gmp, p11-kit, libtasn1, perl, psmisc, time, vcftools, rWrapper, rPackages}: +{ stdenv, callPackage, buildPerlPackage, fetchurl, fetchFromGitHub, perlPackages, R, bwa, samtools, pkgconfig, zlib, htslib, curl, bzip2, lzma, gnutls, nettle, gmp, p11-kit, libtasn1, perl, psmisc, time, vcftools, rWrapper, rPackages }: let ascat = fetchurl { @@ -98,7 +98,7 @@ let url = "mirror://cpan/authors/id/C/CD/CDRAUG/${name}.tar.gz"; sha256 = "0yvhgifs8g9rwdcq84zw4b005nq2jml6c75zgjscv6d2pd3lj1ss"; }; - propagatedBuildInputs = with perlPackages; [DBI DataStag Error GD Graph HTTPMessage HTTPMessage IOstringy IOString IPCRun LWP ListMoreUtils SetScalar TestMost TestRequiresInternet URI XMLDOM XMLDOMXPath XMLLibXML XMLLibXML libxml_perl XMLSAX XMLSAXBase XMLSAXWriter XMLTwig XMLWriter YAML DBFile ]; + propagatedBuildInputs = with perlPackages; [ DBI DataStag Error GD Graph HTTPMessage HTTPMessage IOstringy IOString IPCRun LWP ListMoreUtils SetScalar TestMost TestRequiresInternet URI XMLDOM XMLDOMXPath XMLLibXML XMLLibXML libxml_perl XMLSAX XMLSAXBase XMLSAXWriter XMLTwig XMLWriter YAML DBFile ]; nativeBuildInputs = with perlPackages; [ TestException TestWarn TestDifferences TestDeep ]; }; @@ -108,7 +108,7 @@ let url = "mirror://cpan/authors/id/C/CM/CMUNGALL/${name}.tar.gz"; sha256 = "0ncf4l39ka23nb01jlm6rzxdb5pqbip01x0m38bnvf1gim825caa"; }; - propagatedBuildInputs = with perlPackages; [IOString Graph XMLLibXSLT ]; + propagatedBuildInputs = with perlPackages; [ IOString Graph XMLLibXSLT ]; }; XMLDOMXPath = buildPerlPackage rec { @@ -117,7 +117,7 @@ let url = "mirror://cpan/authors/id/M/MI/MIROD/${name}.tar.gz"; sha256 = "1si9m1pqih3ibbd6jnw69fh98dd4krxpx90p65x9j4aja55afwq1"; }; - propagatedBuildInputs= with perlPackages; [XMLDOM XMLXPathEngine]; + propagatedBuildInputs = with perlPackages; [ XMLDOM XMLXPathEngine ]; doCheck = false; }; @@ -215,10 +215,10 @@ let name = "alleleCount.pl-${version}"; version = "4.0.1"; src = fetchFromGitHub { - owner = "cancerit"; - repo = "alleleCount"; - rev = "v${version}"; - sha256 = "0nkwnjqglgshzhlmz1r0khdjai9mfz4ih8bzrzg0g18d1725k6gp"; + owner = "cancerit"; + repo = "alleleCount"; + rev = "v${version}"; + sha256 = "0nkwnjqglgshzhlmz1r0khdjai9mfz4ih8bzrzg0g18d1725k6gp"; }; preConfigure = '' cd perl @@ -266,4 +266,5 @@ let ]; }; -in ascatNGS +in +ascatNGS diff --git a/tools/ascat-callCNV.nix b/tools/ascat-callCNV.nix index f74bb18..aa1a8ec 100644 --- a/tools/ascat-callCNV.nix +++ b/tools/ascat-callCNV.nix @@ -1,12 +1,12 @@ { bionix , ref , gc -, indexAttrs ? {} -, bamIndexAttrs ? {} +, indexAttrs ? { } +, bamIndexAttrs ? { } , flags ? null }: -{tumour, normal, gender}: +{ tumour, normal, gender }: with bionix; with lib; diff --git a/tools/ascat.nix b/tools/ascat.nix index 700cb4e..142e0d9 100644 --- a/tools/ascat.nix +++ b/tools/ascat.nix @@ -1,17 +1,17 @@ -{bionix}: +{ bionix }: with bionix; { - app = pkgs.callPackage ./ascat-app.nix {}; + app = pkgs.callPackage ./ascat-app.nix { }; /* Generate GC correction file for ascatNGS. - Type: gccorrect :: {ref :: fasta, chrPrefix :: string, ...} -> (snps :: VCF) -> (gc :: GC) + Type: gccorrect :: {ref :: fasta, chrPrefix :: string, ...} -> (snps :: VCF) -> (gc :: GC) */ gccorrect = callBionixE ./ascat-gccorrect.nix; /* Call CNVs using ascatNGS. Gender is a string as per ascatNGS docs (e.g., "XX"). - Type: callCNV :: {ref :: fasta, gc :: GC, ...} -> {tumour :: bam, normal :: bam, gender :: string} -> CNVs + Type: callCNV :: {ref :: fasta, gc :: GC, ...} -> {tumour :: bam, normal :: bam, gender :: string} -> CNVs */ callCNV = callBionixE ./ascat-callCNV.nix; } diff --git a/tools/bowtie-align.nix b/tools/bowtie-align.nix index c83c49c..8df3dba 100644 --- a/tools/bowtie-align.nix +++ b/tools/bowtie-align.nix @@ -2,7 +2,7 @@ , ref , bamOutput ? true , flags ? null -, indexAttrs ? {} +, indexAttrs ? { } }: { input1 @@ -18,7 +18,8 @@ let fa = f: matchFiletype "bowtie2-ref" { fa = _: f; } f; fq = f: matchFiletype "bowtie2-input" { fq = _: f; gz = matchFiletype' "bowtie2-input" { fq = _: f; }; } f; -in stage { +in +stage { name = "bowtie2-align"; buildInputs = with pkgs; [ bowtie2 bc samtools ]; buildCommand = '' @@ -32,6 +33,6 @@ in stage { ${optionalString bamOutput "| samtools view -b"} \ > $out ''; - passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};}; + passthru.filetype = if bamOutput then filetype.bam { inherit ref; sorting = sort.none { }; } else filetype.sam { inherit ref; sorting = sort.name { }; }; passthru.multicore = true; } diff --git a/tools/bowtie.nix b/tools/bowtie.nix index df97d36..2208dff 100644 --- a/tools/bowtie.nix +++ b/tools/bowtie.nix @@ -4,12 +4,12 @@ with bionix; { /* Align a sequence against a reference - Type: align :: {ref :: fasta, bamOutput :: bool, ...} -> {input1 :: fastq, input2 :: fastq} -> bam/sam + Type: align :: {ref :: fasta, bamOutput :: bool, ...} -> {input1 :: fastq, input2 :: fastq} -> bam/sam */ align = callBionixE ./bowtie-align.nix; /* Create a Bowtie index - Type: index :: {seed :: int, ...} -> fasta -> index + Type: index :: {seed :: int, ...} -> fasta -> index */ index = callBionixE ./bowtie-index.nix; } diff --git a/tools/bwa-mem.nix b/tools/bwa-mem.nix index 31c9487..52e73d1 100644 --- a/tools/bwa-mem.nix +++ b/tools/bwa-mem.nix @@ -2,7 +2,7 @@ , ref , bamOutput ? true , flags ? null -, indexAttrs ? {} +, indexAttrs ? { } }: { input1 @@ -18,7 +18,8 @@ let fa = f: matchFiletype "bwa-ref" { fa = _: f; } f; fq = f: matchFiletype "bwa-input" { fq = _: f; gz = matchFiletype' "bwa-input" { fq = _: f; }; } f; -in stage { +in +stage { name = "bwa-mem"; buildInputs = with pkgs; [ bwa bc ] ++ optional bamOutput samtools; buildCommand = '' @@ -35,6 +36,6 @@ in stage { ${optionalString bamOutput "| samtools view -b"} \ > $out ''; - passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};}; + passthru.filetype = if bamOutput then filetype.bam { inherit ref; sorting = sort.none { }; } else filetype.sam { inherit ref; sorting = sort.name { }; }; passthru.multicore = true; } diff --git a/tools/bwa-mem2-app.nix b/tools/bwa-mem2-app.nix index 49f4af3..6ee2ab4 100644 --- a/tools/bwa-mem2-app.nix +++ b/tools/bwa-mem2-app.nix @@ -1,5 +1,5 @@ { clangStdenv, fetchFromGitHub, zlib }: - clangStdenv.mkDerivation rec { +clangStdenv.mkDerivation rec { pname = "bwa-mem2"; version = "2.1"; src = fetchFromGitHub { diff --git a/tools/bwa-mem2.nix b/tools/bwa-mem2.nix index b16b7ea..843c5b0 100644 --- a/tools/bwa-mem2.nix +++ b/tools/bwa-mem2.nix @@ -2,7 +2,7 @@ , ref , bamOutput ? true , flags ? null -, indexAttrs ? {} +, indexAttrs ? { } }: { input1 @@ -18,7 +18,8 @@ let fa = f: matchFiletype "bwa-ref" { fa = _: f; } f; fq = f: matchFiletype "bwa-input" { fq = _: f; gz = matchFiletype' "bwa-input" { fq = _: f; }; } f; -in stage { +in +stage { name = "bwa-mem2"; buildInputs = with pkgs; [ bionix.bwa.app2 bc ] ++ optional bamOutput samtools; buildCommand = '' @@ -35,6 +36,6 @@ in stage { ${optionalString bamOutput "| samtools view -b"} \ > $out ''; - passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};}; + passthru.filetype = if bamOutput then filetype.bam { inherit ref; sorting = sort.none { }; } else filetype.sam { inherit ref; sorting = sort.name { }; }; passthru.multicore = true; } diff --git a/tools/bwa.nix b/tools/bwa.nix index 90a8bd3..6fb7d71 100644 --- a/tools/bwa.nix +++ b/tools/bwa.nix @@ -3,18 +3,18 @@ with bionix; rec { - app2 = pkgs.callPackage ./bwa-mem2-app.nix {}; + app2 = pkgs.callPackage ./bwa-mem2-app.nix { }; /* Align read against a reference: defaults to bwa-mem */ align = bionix.bwa.mem; /* Align reads against a reference using bwa-mem - Type: bwa-mem :: {ref :: fasta, bamOutput :: bool, ...} -> {input1, input2} -> bam/sam + Type: bwa-mem :: {ref :: fasta, bamOutput :: bool, ...} -> {input1, input2} -> bam/sam */ mem = callBionixE ./bwa-mem.nix; mem2 = callBionixE ./bwa-mem2.nix; /* Creates an reference index for BWA - Type: index :: {...} -> fasta -> BWA index + Type: index :: {...} -> fasta -> BWA index */ index = callBionixE ./bwa-index.nix; index2 = callBionixE ./bwa-index2.nix; diff --git a/tools/cnvkit-batch.nix b/tools/cnvkit-batch.nix index 113e0c9..7f7893f 100644 --- a/tools/cnvkit-batch.nix +++ b/tools/cnvkit-batch.nix @@ -1,17 +1,18 @@ -{bionix -,targets ? null -,annotations ? null -,flags ? null -,indexAttrs ? {}}: +{ bionix +, targets ? null +, annotations ? null +, flags ? null +, indexAttrs ? { } +}: -{normals ? [], tumours}: +{ normals ? [ ], tumours }: with bionix; with lib; with types; let - getref = f: matchFiletype "cnvkit-batch" { bam = {ref, ...}: ref; } f; + getref = matchFiletype "cnvkit-batch" { bam = { ref, ... }: ref; }; refs = map getref normals ++ map getref tumours; ref = head refs; sorted = matchFileSorting "cnvkit-batch" { coord = _: true; }; diff --git a/tools/cnvkit-scatter.nix b/tools/cnvkit-scatter.nix index 2899584..6cd5b96 100644 --- a/tools/cnvkit-scatter.nix +++ b/tools/cnvkit-scatter.nix @@ -1,5 +1,6 @@ -{bionix -,flags ? null}: +{ bionix +, flags ? null +}: input: diff --git a/tools/cnvkit.nix b/tools/cnvkit.nix index fc55d94..812652a 100644 --- a/tools/cnvkit.nix +++ b/tools/cnvkit.nix @@ -1,15 +1,15 @@ -{bionix}: +{ bionix }: with bionix; { /* Call CNVs - Type: callCNV :: {targets :: target file, annotations :: annotation file, ...} -> {normals :: [bam], tumours :: [bam]} -> CNVs + Type: callCNV :: {targets :: target file, annotations :: annotation file, ...} -> {normals :: [bam], tumours :: [bam]} -> CNVs */ callCNV = callBionixE ./cnvkit-batch.nix; /* Scatter plot from CNV calls - Type: scatterPlot :: {} -> CNVs -> PDF + Type: scatterPlot :: {} -> CNVs -> PDF */ scatterPlot = callBionixE ./cnvkit-scatter.nix; } diff --git a/tools/compression-bunzip2.nix b/tools/compression-bunzip2.nix index 06fe682..a36b66d 100644 --- a/tools/compression-bunzip2.nix +++ b/tools/compression-bunzip2.nix @@ -1,4 +1,4 @@ -{bionix}: +{ bionix }: with bionix; diff --git a/tools/compression-bzip2.nix b/tools/compression-bzip2.nix index 6dfb749..5332e7d 100644 --- a/tools/compression-bzip2.nix +++ b/tools/compression-bzip2.nix @@ -1,4 +1,4 @@ -{bionix}: +{ bionix }: with bionix; diff --git a/tools/compression-gunzip.nix b/tools/compression-gunzip.nix index 6c98f91..96a4e49 100644 --- a/tools/compression-gunzip.nix +++ b/tools/compression-gunzip.nix @@ -1,4 +1,4 @@ -{bionix}: +{ bionix }: with bionix; diff --git a/tools/compression-gzip.nix b/tools/compression-gzip.nix index 50fca9e..0229b94 100644 --- a/tools/compression-gzip.nix +++ b/tools/compression-gzip.nix @@ -1,4 +1,4 @@ -{bionix}: +{ bionix }: with bionix; diff --git a/tools/compression.nix b/tools/compression.nix index bfcc292..c9c8777 100644 --- a/tools/compression.nix +++ b/tools/compression.nix @@ -1,4 +1,4 @@ -{bionix}: +{ bionix }: with bionix; with types; @@ -8,42 +8,51 @@ let gunzip = callBionixE ./compression-gunzip.nix; bunzip2 = callBionixE ./compression-bunzip2.nix; -in { - uncompress = attrs: f: matchFiletype "uncompress" { - fa = _: f; - fq = _: f; - bam = _: f; - sam = _: f; - cram = _: f; - vcf = _: f; - bed = _: f; - gz = _: gunzip attrs f; - bz2 = _: bunzip2 attrs f; - } f; +in +{ + uncompress = attrs: f: matchFiletype "uncompress" + { + fa = _: f; + fq = _: f; + bam = _: f; + sam = _: f; + cram = _: f; + vcf = _: f; + bed = _: f; + gz = _: gunzip attrs f; + bz2 = _: bunzip2 attrs f; + } + f; gzip = attrs: f: let gz = gzip attrs f; - in types.matchFiletype "compressed" { - fa = _: gz; - fq = _: gz; - bam = _: gz; - sam = _: gz; - cram = _: gz; - vcf = _: gz; - bed = _: gz; - gz = x: x; - } f; + in + types.matchFiletype "compressed" + { + fa = _: gz; + fq = _: gz; + bam = _: gz; + sam = _: gz; + cram = _: gz; + vcf = _: gz; + bed = _: gz; + gz = x: x; + } + f; bzip2 = attrs: f: let bz2 = bzip2 attrs f; - in types.matchFiletype "compressed" { - fa = _: gz; - fq = _: gz; - bam = _: gz; - sam = _: gz; - cram = _: gz; - vcf = _: gz; - bed = _: gz; - bz2 = x: x; - } f; + in + types.matchFiletype "compressed" + { + fa = _: gz; + fq = _: gz; + bam = _: gz; + sam = _: gz; + cram = _: gz; + vcf = _: gz; + bed = _: gz; + bz2 = x: x; + } + f; } diff --git a/tools/crumble-app.nix b/tools/crumble-app.nix index fc30d5a..d8ed1a7 100644 --- a/tools/crumble-app.nix +++ b/tools/crumble-app.nix @@ -1,4 +1,4 @@ -{stdenv, fetchFromGitHub, autoreconfHook, htslib}: +{ stdenv, fetchFromGitHub, autoreconfHook, htslib }: stdenv.mkDerivation rec { name = "crumble-${version}"; diff --git a/tools/crumble.nix b/tools/crumble.nix index 4256f37..dfb4e31 100644 --- a/tools/crumble.nix +++ b/tools/crumble.nix @@ -3,5 +3,5 @@ with bionix; { - crumble = pkgs.callPackage ./crumble-app.nix {}; + crumble = pkgs.callPackage ./crumble-app.nix { }; } diff --git a/tools/delly-call.nix b/tools/delly-call.nix index 5378810..80084be 100644 --- a/tools/delly-call.nix +++ b/tools/delly-call.nix @@ -1,4 +1,4 @@ -{ bionix, flags ? "", faidxAttrs ? {}, indexAttrs ? {}}: +{ bionix, flags ? "", faidxAttrs ? { }, indexAttrs ? { } }: inputs: @@ -41,6 +41,6 @@ stage { ''; passthru = { multicore = true; - filetype = filetype.vcf { ref = ref; }; + filetype = filetype.vcf { inherit ref; }; }; } diff --git a/tools/delly.nix b/tools/delly.nix index d94457c..22d4d89 100644 --- a/tools/delly.nix +++ b/tools/delly.nix @@ -4,7 +4,7 @@ with bionix; { /* Call structural variants - Type: call { ... } -> [bam] -> vcf + Type: call { ... } -> [bam] -> vcf */ call = callBionixE ./delly-call.nix; } diff --git a/tools/facets-app.nix b/tools/facets-app.nix index 6553a67..fe2a31b 100644 --- a/tools/facets-app.nix +++ b/tools/facets-app.nix @@ -1,8 +1,9 @@ -{buildRPackage -,fetchFromGitHub -,R -,htslib -,zlib}: +{ buildRPackage +, fetchFromGitHub +, R +, htslib +, zlib +}: let pctGCdata = buildRPackage rec { @@ -18,7 +19,8 @@ let buildInputs = [ R ]; }; -in buildRPackage rec{ +in +buildRPackage rec{ name = "facets-${version}"; version = "0.6.1"; requireX = false; diff --git a/tools/facets-call.nix b/tools/facets-call.nix index a736627..43c4c7f 100644 --- a/tools/facets-call.nix +++ b/tools/facets-call.nix @@ -1,6 +1,6 @@ -{bionix}: +{ bionix }: -{vcf, bams}: +{ vcf, bams }: with bionix; with types; diff --git a/tools/facets.nix b/tools/facets.nix index 14110a9..b91e422 100644 --- a/tools/facets.nix +++ b/tools/facets.nix @@ -1,12 +1,12 @@ -{bionix}: +{ bionix }: with bionix; { - app = lib.callPackageWith (pkgs // pkgs.rPackages) ./facets-app.nix {}; + app = lib.callPackageWith (pkgs // pkgs.rPackages) ./facets-app.nix { }; /* Call CNVs - Type: callCnv :: {...} -> {vcf, bams :: [bams]} -> CNVs + Type: callCnv :: {...} -> {vcf, bams :: [bams]} -> CNVs */ callCNV = callBionixE ./facets-call.nix; } diff --git a/tools/fastp-check.nix b/tools/fastp-check.nix index 06f00ae..9467cd9 100644 --- a/tools/fastp-check.nix +++ b/tools/fastp-check.nix @@ -1,10 +1,10 @@ { bionix , flags ? null -} : +}: { input1 , input2 ? null -} : +}: with bionix; with lib; @@ -15,26 +15,27 @@ let out = stage { - name = "fastp"; - buildInputs = [ pkgs.fastp ]; - outputs = [ "out" "fastq1" "json" ] ++ (if input2 != null then [ "fastq2" ] else []); - buildCommand = '' - mkdir -p $out - fastp \ - ${optionalString (flags != null) flags} \ - -i ${fq input1} \ - -o fastq1.fq.gz \ - ${optionalString (input2 != null) '' - -I ${fq input2} \ - -O fastq2.fq.gz \ + name = "fastp"; + buildInputs = [ pkgs.fastp ]; + outputs = [ "out" "fastq1" "json" ] ++ (if input2 != null then [ "fastq2" ] else [ ]); + buildCommand = '' + mkdir -p $out + fastp \ + ${optionalString (flags != null) flags} \ + -i ${fq input1} \ + -o fastq1.fq.gz \ + ${optionalString (input2 != null) '' + -I ${fq input2} \ + -O fastq2.fq.gz \ - cp fastq2.fq.gz $fastq2 - ''} + cp fastq2.fq.gz $fastq2 + ''} - cp fastq1.fq.gz $fastq1 - cp fastp.html $out - cp fastp.json $json - ''; + cp fastq1.fq.gz $fastq1 + cp fastp.html $out + cp fastp.json $json + ''; }; - fqgz = { filetype = filetype.gz (filetype.fastq {}); }; -in out // { fastq1 = out.fastq1 // fqgz; } // (if input2 != null then {fastq2 = out.fastq2 // fqgz; } else {}) + fqgz = { filetype = filetype.gz (filetype.fastq { }); }; +in +out // { fastq1 = out.fastq1 // fqgz; } // (if input2 != null then { fastq2 = out.fastq2 // fqgz; } else { }) diff --git a/tools/fastp.nix b/tools/fastp.nix index 36bc690..454d100 100644 --- a/tools/fastp.nix +++ b/tools/fastp.nix @@ -3,9 +3,9 @@ with bionix; rec { - /* Check and filter fastqs + /* Check and filter fastqs Type: { ... } -> { input1, input2 ? null } -> { out :: html, fastq1 :: fastq, fastq2 :: fastq, json :: JSON } - */ - check = callBionixE ./fastp-check.nix; + */ + check = callBionixE ./fastp-check.nix; } diff --git a/tools/fastqc.nix b/tools/fastqc.nix index 6279682..8c448b9 100644 --- a/tools/fastqc.nix +++ b/tools/fastqc.nix @@ -3,10 +3,10 @@ with bionix; { - fastqc = pkgs.callPackage ./fastqc-app.nix {}; + fastqc = pkgs.callPackage ./fastqc-app.nix { }; /* QC check - Type: check :: {...} -> input :: fastq -> report + Type: check :: {...} -> input :: fastq -> report */ check = callBionixE ./fastqc-check.nix; } diff --git a/tools/gridss-assemble.nix b/tools/gridss-assemble.nix index 1a46eae..15e5ca6 100644 --- a/tools/gridss-assemble.nix +++ b/tools/gridss-assemble.nix @@ -1,12 +1,12 @@ { bionix -, bwaIndexAttrs ? {} -, faidxAttrs ? {} -, indexAttrs ? {} -, collectMetricsAttrs ? {} +, bwaIndexAttrs ? { } +, faidxAttrs ? { } +, indexAttrs ? { } +, collectMetricsAttrs ? { } , flags ? null , config ? null , heapSize ? "31g" -, workdirs ? [] +, workdirs ? [ ] , jobIndex ? null , jobNodes ? null }: @@ -36,7 +36,7 @@ let in assert (all sorted inputs); -assert (homoRef); +assert homoRef; stage rec { name = "gridss-assemble"; @@ -70,6 +70,6 @@ stage rec { touch $out cp -r out.gridss.working $work ''; - passthru.filetype = filetype.bam { ref = ref; sorting = sort.none {}; }; + passthru.filetype = filetype.bam { inherit ref; sorting = sort.none { }; }; passthru.multicore = true; } diff --git a/tools/gridss-callVariants.nix b/tools/gridss-callVariants.nix index f42e3f3..ad659de 100644 --- a/tools/gridss-callVariants.nix +++ b/tools/gridss-callVariants.nix @@ -1,7 +1,7 @@ { bionix , blacklist ? null -, bwaIndexAttrs ? {} -, faidxAttrs ? {} +, bwaIndexAttrs ? { } +, faidxAttrs ? { } , flags ? null , config ? null , heapSize ? "31g" @@ -32,22 +32,22 @@ stage rec { done mkdir $out java -ea -Xmx${heapSize} \ - -Dreference_fasta="ref.fa" \ - -Dsamjdk.create_index=true \ - -Dsamjdk.use_async_io_read_samtools=true \ - -Dsamjdk.use_async_io_write_samtools=true \ - -Dsamjdk.use_async_io_write_tribble=true \ - -Dgridss.gridss.output_to_temp_file=true \ - -cp ${bionix.gridss.jar} gridss.CallVariants \ + -Dreference_fasta="ref.fa" \ + -Dsamjdk.create_index=true \ + -Dsamjdk.use_async_io_read_samtools=true \ + -Dsamjdk.use_async_io_write_samtools=true \ + -Dsamjdk.use_async_io_write_tribble=true \ + -Dgridss.gridss.output_to_temp_file=true \ + -cp ${bionix.gridss.jar} gridss.CallVariants \ VERBOSITY=WARNING \ WORKER_THREADS=$NIX_BUILD_CORES \ - TMP_DIR=. \ - WORKING_DIR=. \ + TMP_DIR=. \ + WORKING_DIR=. \ ${optionalString (config != null) ("OPTIONS_FILE=" + bionix.gridss.gridssConfig config)} \ - REFERENCE_SEQUENCE="ref.fa" \ + REFERENCE_SEQUENCE="ref.fa" \ ${concatMapStringsSep " " (i: "INPUT=\"${i}\"") inputs} \ - OUTPUT="$out/gridss.vcf" \ - ASSEMBLY="$out/gridss.bam" \ + OUTPUT="$out/gridss.vcf" \ + ASSEMBLY="$out/gridss.bam" \ ${optionalString (blacklist != null) ("BLACKLIST=" + blacklist)} \ ${optionalString (flags != null) flags} diff --git a/tools/gridss-computeSamTags.nix b/tools/gridss-computeSamTags.nix index de44478..e07f6a5 100644 --- a/tools/gridss-computeSamTags.nix +++ b/tools/gridss-computeSamTags.nix @@ -1,5 +1,10 @@ -{ bionix, bwaIndexAttrs ? { }, faidxAttrs ? { }, flags ? null, config ? null -, heapSize ? "1G" }: +{ bionix +, bwaIndexAttrs ? { } +, faidxAttrs ? { } +, flags ? null +, config ? null +, heapSize ? "1G" +}: with bionix; with lib; @@ -11,7 +16,8 @@ let ref = matchFiletype "gridss-computeSamTags" { bam = x: x.ref; } input; sorted = matchFileSorting "gridss-computeSamTags" { name = _: true; } input; -in assert (sorted); +in +assert sorted; stage rec { name = "gridss-computeSamTags"; diff --git a/tools/gridss-configFile.nix b/tools/gridss-configFile.nix index d6b600e..3d0d3f8 100644 --- a/tools/gridss-configFile.nix +++ b/tools/gridss-configFile.nix @@ -1,37 +1,41 @@ -{bionix}: +{ bionix }: with bionix; with lib; let - attrsToGridssConfigString = attrsToGridssConfigStringPrepend ""; + attrsToGridssConfigString = attrsToGridssConfigStringPrepend ""; - attrsToGridssConfigStringPrepend = prepend: attrs: - concatStringsSep "\n" ( - attrValues ( - mapAttrs - (name: attr: prepend + (iniLine name attr)) - attrs)); + attrsToGridssConfigStringPrepend = prepend: attrs: + concatStringsSep "\n" ( + attrValues ( + mapAttrs + (name: attr: prepend + (iniLine name attr)) + attrs)); - iniLine = name: attr: - let attrType = builtins.typeOf attr; - in - if (iniLineByAttrType ? ${attrType}) - then (iniLineByAttrType.${attrType} name attr) - else builtins.throw ( - "`gridssConfig` cannot convert attribute of type \"" + attrType + "\"."); + iniLine = name: attr: + let attrType = builtins.typeOf attr; + in + if (iniLineByAttrType ? "${attrType}") + then (iniLineByAttrType."${attrType}" name attr) + else + builtins.throw ( + "`gridssConfig` cannot convert attribute of type \"" + attrType + "\"." + ); - iniLineByAttrType = { - string = name: attr: name + " = " + attr; - int = name: attr: name + " = " + builtins.toString attr; - float = name: attr: name + " = " + ( - builtins.head ( - builtins.match "([0-9]+\.0?[1-9]*)0+" (builtins.toString attr))); - bool = name: attr: name + " = " + (if attr == true then "true" else "false"); - set = name: attr: attrsToGridssConfigStringPrepend (name + ".") attr; - # Allows for repeated fields (e.g. for adapters): - list = name: attr: concatStringsSep "\n" (map (x: iniLine name x) attr); - }; -in configAttrs: (pkgs.writeText - "gridss.properties.override" - ((attrsToGridssConfigString configAttrs) + "\n")) + iniLineByAttrType = { + string = name: attr: name + " = " + attr; + int = name: attr: name + " = " + builtins.toString attr; + float = name: attr: name + " = " + ( + builtins.head ( + builtins.match "([0-9]+\.0?[1-9]*)0+" (builtins.toString attr) + )); + bool = name: attr: name + " = " + (if attr then "true" else "false"); + set = name: attrsToGridssConfigStringPrepend (name + "."); + # Allows for repeated fields (e.g. for adapters): + list = name: attr: concatStringsSep "\n" (map (iniLine name) attr); + }; +in +configAttrs: (pkgs.writeText + "gridss.properties.override" + ((attrsToGridssConfigString configAttrs) + "\n")) diff --git a/tools/gridss-extractSVReads.nix b/tools/gridss-extractSVReads.nix index 6691808..ea268cf 100644 --- a/tools/gridss-extractSVReads.nix +++ b/tools/gridss-extractSVReads.nix @@ -1,10 +1,10 @@ { bionix -, dictIndexAttrs ? {} -, faidxAttrs ? {} +, dictIndexAttrs ? { } +, faidxAttrs ? { } , flags ? null , unmappedReads ? false , minClipLength ? 5 -, collectMetricsAttrs ? {} +, collectMetricsAttrs ? { } , config ? null }: diff --git a/tools/gridss-softClipsToSplitReads.nix b/tools/gridss-softClipsToSplitReads.nix index 93d02df..ef53322 100644 --- a/tools/gridss-softClipsToSplitReads.nix +++ b/tools/gridss-softClipsToSplitReads.nix @@ -1,6 +1,6 @@ { bionix -, bwaIndexAttrs ? {} -, faidxAttrs ? {} +, bwaIndexAttrs ? { } +, faidxAttrs ? { } , alignerStreaming ? false , flags ? null , config ? null @@ -37,7 +37,7 @@ stage rec { ${optionalString (config != null) ("OPTIONS_FILE=" + bionix.gridss.gridssConfig config)} \ WORKER_THREADS=$NIX_BUILD_CORES \ ${optionalString (flags != null) flags} - ''; - passthru.filetype = filetype.bam { ref = ref; sorting = sort.none {}; }; + ''; + passthru.filetype = filetype.bam { inherit ref; sorting = sort.none { }; }; passthru.multicore = true; } diff --git a/tools/gridss-variants.nix b/tools/gridss-variants.nix index a5c470a..62a89f6 100644 --- a/tools/gridss-variants.nix +++ b/tools/gridss-variants.nix @@ -1,7 +1,14 @@ -{ bionix, bwaIndexAttrs ? { }, faidxAttrs ? { }, indexAttrs ? { } -, assemblyAttrs ? { }, collectMetricsAttrs ? { } +{ bionix +, bwaIndexAttrs ? { } +, faidxAttrs ? { } +, indexAttrs ? { } +, assemblyAttrs ? { } +, collectMetricsAttrs ? { } , softClipsToSplitReadsAttrs ? { flags = "REALIGN_ENTIRE_READ=true"; } -, config ? null, heapSize ? "4g", shards ? 10 }: +, config ? null +, heapSize ? "4g" +, shards ? 10 +}: with bionix; with lib; @@ -47,15 +54,16 @@ let for f in ${bionix.bwa.index bwaIndexAttrs ref}/*; do ln -s $f done - ${concatMapStringsSep "\n" (linkSV) inputs} + ${concatMapStringsSep "\n" linkSV inputs} ${linkSV assembly} ${concatMapStringsSep "\n" (linkInput collectMetrics collectMetricsAttrs) inputs} ${linkInput collectMetrics collectMetricsAttrs assembly} ''; -in assert (all sorted inputs); -assert (homoRef); +in +assert (all sorted inputs); +assert homoRef; rec { identify = stage rec { @@ -80,7 +88,7 @@ rec { mv out.vcf $out ''; passthru = { - filetype = filetype.vcf { ref = ref; }; + filetype = filetype.vcf { inherit ref; }; gridss.assembly = assembly; multicore = true; }; @@ -115,7 +123,7 @@ rec { mv out.vcf $out ''; passthru = { - filetype = filetype.vcf { ref = ref; }; + filetype = filetype.vcf { inherit ref; }; gridss.assembly = assembly; multicore = true; }; diff --git a/tools/gridss.nix b/tools/gridss.nix index 7372864..db577ad 100644 --- a/tools/gridss.nix +++ b/tools/gridss.nix @@ -10,60 +10,65 @@ rec { }; /* Generate configuration file for GRIDSS. Takes attribute sets to GRIDSS ini style format. - Type: genConfig :: attrSet -> ini file + Type: genConfig :: attrSet -> ini file */ genConfig = callBionix ./gridss-configFile.nix { }; /* Invoke the callVariants tool - Type: callVariants :: {blacklist :: drv = null, config :: ini = null, heapSize :: String = "31g", ...} -> [bam] -> variants + Type: callVariants :: {blacklist :: drv = null, config :: ini = null, heapSize :: String = "31g", ...} -> [bam] -> variants */ callVariants = callBionixE ./gridss-callVariants.nix; /* Invoke computeSamTags tool - Type: computeSamTags :: {config :: ini = null, heapSize :: String = "1G", ...} -> bam -> bam + Type: computeSamTags :: {config :: ini = null, heapSize :: String = "1G", ...} -> bam -> bam */ computeSamTags = callBionixE ./gridss-computeSamTags.nix; /* Invoke softClipsToSplitReads tool - Type: softClipsToSplitReads :: {alignerStreaming :: Bool = false, config :: ini = null, heapSize :: String = "2G", ...} -> bam -> bam + Type: softClipsToSplitReads :: {alignerStreaming :: Bool = false, config :: ini = null, heapSize :: String = "2G", ...} -> bam -> bam */ softClipsToSplitReads = callBionixE ./gridss-softClipsToSplitReads.nix; /* Invoke collectMetrics tool - Type: collectMetrics :: {thresholdCoverage :: Int = 10000, config :: ini = null, heapSize :: String = "1G", ...} -> bam -> metrics + Type: collectMetrics :: {thresholdCoverage :: Int = 10000, config :: ini = null, heapSize :: String = "1G", ...} -> bam -> metrics */ collectMetrics = callBionixE ./gridss-collectMetrics.nix; /* Invoke extractSVReads tool - Type: extractSVReads :: {unmappedReads :: Bool = false, minClipLength :: Int = 5, config :: ini = null, ...} -> bam -> bam + Type: extractSVReads :: {unmappedReads :: Bool = false, minClipLength :: Int = 5, config :: ini = null, ...} -> bam -> bam */ extractSVReads = callBionixE ./gridss-extractSVReads.nix; /* Invoke assembly tool - Type: assemble :: {config :: ini = null, heapSize :: String = "31g", ...} -> [bam] -> bam + Type: assemble :: {config :: ini = null, heapSize :: String = "31g", ...} -> [bam] -> bam */ assemble = callBionixE ./gridss-assemble.nix; shardedAssemble = n: a: input: let - assemblies = genList (i: - bionix.gridss.assemble (a // { - jobNodes = n; - jobIndex = i; - }) input) n; - in if n <= 1 then + assemblies = genList + (i: + bionix.gridss.assemble + (a // { + jobNodes = n; + jobIndex = i; + }) + input) + n; + in + if n <= 1 then bionix.gridss.assemble a input else bionix.gridss.assemble (a // { workdirs = map (a: a.work) assemblies; }) - input; + input; /* Invoke identifyVariants tool - Type: identifyVariants :: {config :: ini = null, heapSize :: String = "4g", ...} -> [bam] -> VCF + Type: identifyVariants :: {config :: ini = null, heapSize :: String = "4g", ...} -> [bam] -> VCF */ identifyVariants = exec (attrs: input: ((callBionix ./gridss-variants.nix attrs) input).identify); /* Invoke annotateVariants tool - Type: annotateVariants :: {config :: ini = null, heapSize :: String = "4g", ...} -> [bam] -> VCF + Type: annotateVariants :: {config :: ini = null, heapSize :: String = "4g", ...} -> [bam] -> VCF */ annotateVariants = exec (attrs: input: ((callBionix ./gridss-variants.nix attrs) input).annotate); @@ -73,7 +78,7 @@ rec { ((callBionix ./gridss-variants.nix attrs) input).annotateAndAssemble); /* Preprocess BAM files to extract SV reads - Type: preprocessBam :: bam -> bam + Type: preprocessBam :: bam -> bam */ preprocessBam = with samtools; flip pipe [ @@ -84,7 +89,7 @@ rec { ]; /* Call SVs: entire pipeline including preprocessing. It is recommended to use this function rather than the individual above tools. - Type: [bam] -> GRIDSS result + Type: [bam] -> GRIDSS result */ call = inputs: gridss.annotateVariants { } (map gridss.preprocessBam inputs); diff --git a/tools/hisat2-align.nix b/tools/hisat2-align.nix index 7a216a6..0a1eb25 100644 --- a/tools/hisat2-align.nix +++ b/tools/hisat2-align.nix @@ -2,7 +2,7 @@ , ref , bamOutput ? true , flags ? null -, indexAttrs ? {} +, indexAttrs ? { } }: { input1 @@ -18,7 +18,8 @@ let fa = f: matchFiletype "hisat2-ref" { fa = _: f; } f; fq = f: matchFiletype "hisat2-input" { fq = _: f; gz = matchFiletype' "hisat2-input" { fq = _: "<(gunzip < ${f})"; }; } f; -in stage { +in +stage { name = "hisat2-align"; buildInputs = with pkgs; [ hisat2 bc samtools ]; buildCommand = '' @@ -33,6 +34,6 @@ in stage { | samtools sort -n \ > $out ''; - passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};}; + passthru.filetype = if bamOutput then filetype.bam { inherit ref; sorting = sort.none { }; } else filetype.sam { inherit ref; sorting = sort.name { }; }; passthru.multicore = true; } diff --git a/tools/hisat2-index.nix b/tools/hisat2-index.nix index ffdb4ed..d9d5cd1 100644 --- a/tools/hisat2-index.nix +++ b/tools/hisat2-index.nix @@ -1,8 +1,8 @@ { bionix , gtf ? null , flags ? null -, extractSpliceSitesAttrs ? {} -, extractExonsAttrs ? {} +, extractSpliceSitesAttrs ? { } +, extractExonsAttrs ? { } }: ref: diff --git a/tools/hisat2.nix b/tools/hisat2.nix index 8cc4ddf..e3e9995 100644 --- a/tools/hisat2.nix +++ b/tools/hisat2.nix @@ -4,12 +4,12 @@ with bionix; rec { /* Align read against a reference - Type: hisat2-mem :: {ref :: fasta, bamOutput :: bool, ...} -> {input1, input2} -> bam/sam + Type: hisat2-mem :: {ref :: fasta, bamOutput :: bool, ...} -> {input1, input2} -> bam/sam */ align = callBionixE ./hisat2-align.nix; /* Creates an reference index for HISAT2 - Type: index :: {...} -> fasta -> HISAT2 index + Type: index :: {...} -> fasta -> HISAT2 index */ index = callBionixE ./hisat2-index.nix; diff --git a/tools/infercnv-app.nix b/tools/infercnv-app.nix index 5114e2f..36b9956 100644 --- a/tools/infercnv-app.nix +++ b/tools/infercnv-app.nix @@ -1,21 +1,22 @@ -{stdenv - ,fetchurl - ,fetchFromGitHub - ,callPackage - ,nettools - ,rPackages - ,rWrapper - ,R - ,jre - ,python - ,pythonPackages - ,darwin - ,gettext - ,gfortran -,makeWrapper}: +{ stdenv +, fetchurl +, fetchFromGitHub +, callPackage +, nettools +, rPackages +, rWrapper +, R +, jre +, python +, pythonPackages +, darwin +, gettext +, gfortran +, makeWrapper +}: let - buildRPackage = rPackages.buildRPackage; + inherit (rPackages) buildRPackage; GMD = buildRPackage rec { name = "GMD-${version}"; @@ -59,7 +60,7 @@ let owner = "broadinstitute"; repo = "inferCNV"; rev = "cf442af0db6191fa8ba57c4921ac2d1f98c2c39d"; - sha256 = "0cv8qiaqpd6b4152dplnzvgv77cmk961rmvzr27qgmlaazc5hblh"; + sha256 = "0cv8qiaqpd6b4152dplnzvgv77cmk961rmvzr27qgmlaazc5hblh"; }; propagatedBuildInputs = with rPackages; [ R GMD NGCHMR RColorBrewer gplots optparse logging ]; }; @@ -75,9 +76,10 @@ let sha256 = "1pz710ig8nnydz329ry8fydccbrp3arp614dgba3bcyy9flm3gnw"; }; -in stdenv.mkDerivation rec { - name = inferCNV.name; - src = inferCNV.src; +in +stdenv.mkDerivation rec { + inherit (inferCNV) name; + inherit (inferCNV) src; buildInputs = [ r makeWrapper py ]; propagatedBuildInputs = [ jre ]; installPhase = '' diff --git a/tools/infercnv-infer.nix b/tools/infercnv-infer.nix index 0915691..e499f6c 100644 --- a/tools/infercnv-infer.nix +++ b/tools/infercnv-infer.nix @@ -1,12 +1,14 @@ -{bionix -,flags ? null}: +{ bionix +, flags ? null +}: with bionix; with lib; -{ref -,expr -,pos}: +{ ref +, expr +, pos +}: stage { name = "inferCNV"; diff --git a/tools/infercnv.nix b/tools/infercnv.nix index 71ed274..86ffb74 100644 --- a/tools/infercnv.nix +++ b/tools/infercnv.nix @@ -1,8 +1,8 @@ -{bionix}: +{ bionix }: with bionix; { - app = callPackage ./infercnv-app.nix {}; - infercnv = callBionixE ./infercnv-infer.nix {}; + app = callPackage ./infercnv-app.nix { }; + infercnv = callBionixE ./infercnv-infer.nix { }; } diff --git a/tools/kallisto-index.nix b/tools/kallisto-index.nix index 3b92d38..074a953 100644 --- a/tools/kallisto-index.nix +++ b/tools/kallisto-index.nix @@ -1,6 +1,7 @@ -{bionix +{ bionix , kmerSize ? 31 -, unique ? false}: +, unique ? false +}: with bionix; with lib; diff --git a/tools/kallisto-quant.nix b/tools/kallisto-quant.nix index 3efeaaa..b7481d8 100644 --- a/tools/kallisto-quant.nix +++ b/tools/kallisto-quant.nix @@ -1,5 +1,5 @@ -{bionix -, indexFlags ? {} +{ bionix +, indexFlags ? { } , bias ? false , bootstrapSamples ? 0 , seed ? 42 @@ -10,7 +10,8 @@ , rfStranded ? false , fragmentLength ? null , fragmentSD ? null -, ref}: +, ref +}: with bionix; with lib; @@ -21,7 +22,7 @@ inputs: let inherit (bionix.types) matchFiletype'; - isFastQ = matchFiletype' "kallisto-quant" {fq = _: true; gz = isFastQ; }; + isFastQ = matchFiletype' "kallisto-quant" { fq = _: true; gz = isFastQ; }; empty = ./kallisto-quant-empty.h5; @@ -45,7 +46,7 @@ let in -assert (all (x: isFastQ (x.filetype)) inputs); +assert (all (x: isFastQ x.filetype) inputs); stage { name = "kallisto-quant"; diff --git a/tools/kallisto.nix b/tools/kallisto.nix index 13de61e..e096d11 100644 --- a/tools/kallisto.nix +++ b/tools/kallisto.nix @@ -1,4 +1,4 @@ -{bionix}: +{ bionix }: with bionix; @@ -6,7 +6,7 @@ with bionix; index = callBionixE ./kallisto-index.nix; /* Quantify reads against a transcriptome - Type: { bias :: bool, bootstrapSamples :: int, seed :: int, plaintext :: bool, fusion :: bool, single :: bool, frStranded :: bool, rfStranded :: bool, fragmentLength :: Int, fragmentSD :: real } -> [fastq] -> kallisto + Type: { bias :: bool, bootstrapSamples :: int, seed :: int, plaintext :: bool, fusion :: bool, single :: bool, frStranded :: bool, rfStranded :: bool, fragmentLength :: Int, fragmentSD :: real } -> [fastq] -> kallisto */ quant = callBionixE ./kallisto-quant.nix; } diff --git a/tools/last-align.nix b/tools/last-align.nix index b2c8460..b326877 100644 --- a/tools/last-align.nix +++ b/tools/last-align.nix @@ -1,7 +1,7 @@ { bionix , ref , flags ? null -, indexAttrs ? {} +, indexAttrs ? { } }: { input1 @@ -16,10 +16,11 @@ let fa = f: matchFiletype "last-ref" { fa = _: f; } f; fq = f: matchFiletype "last-input" { fq = _: f; } f; fqfa = f: matchFiletype "last-input" { fq = _: f; fa = _: f; } f; - Q = f: matchFiletype "last-input" { fq = _: 1; fa = _: 0; } f; - parallel = f: matchFiletype "last-input" { fq = _: "parallel-fastq"; fa = _: "parallel-fasta"; } f; + Q = matchFiletype "last-input" { fq = _: 1; fa = _: 0; }; + parallel = matchFiletype "last-input" { fq = _: "parallel-fastq"; fa = _: "parallel-fasta"; }; -in stage { +in +stage { name = "last-align"; buildInputs = with pkgs; [ last ]; buildCommand = '' diff --git a/tools/last.nix b/tools/last.nix index 3755bde..870bfe6 100644 --- a/tools/last.nix +++ b/tools/last.nix @@ -4,7 +4,7 @@ with bionix; rec { /* Align reads against a reference genome - Type: { ref :: fastq, ... } -> { input1, input2} -> last + Type: { ref :: fastq, ... } -> { input1, input2} -> last */ align = callBionixE ./last-align.nix; index = callBionixE ./last-index.nix; diff --git a/tools/lumpy-call.nix b/tools/lumpy-call.nix index fffc329..56ac650 100644 --- a/tools/lumpy-call.nix +++ b/tools/lumpy-call.nix @@ -21,7 +21,8 @@ let ''; }; -in assert (length (unique refs) == 1); +in +assert (length (unique refs) == 1); stage { name = "lumpy"; @@ -36,5 +37,5 @@ stage { ${flags} \ -o $out ''; - passthru.filetype = filetype.vcf { ref = ref; }; + passthru.filetype = filetype.vcf { inherit ref; }; } diff --git a/tools/lumpy.nix b/tools/lumpy.nix index f82134e..59c968f 100644 --- a/tools/lumpy.nix +++ b/tools/lumpy.nix @@ -4,7 +4,7 @@ with bionix; { /* Call structural variants - Type: { ... } -> [bam] -> vcf + Type: { ... } -> [bam] -> vcf */ call = callBionixE ./lumpy-call.nix; } diff --git a/tools/manta-call.nix b/tools/manta-call.nix index 099d5e6..21cbeba 100644 --- a/tools/manta-call.nix +++ b/tools/manta-call.nix @@ -1,6 +1,6 @@ -{ bionix, indexAttrs ? {}, faidxAttrs ? {}, flags ? "" }: +{ bionix, indexAttrs ? { }, faidxAttrs ? { }, flags ? "" }: -{ normals ? [], tumour ? null }: +{ normals ? [ ], tumour ? null }: with bionix; with lib; @@ -8,7 +8,7 @@ with types; let getref = matchFiletype "manta-call" { bam = x: x.ref; }; - refs = map getref normals ++ optionals (tumour != null) [(getref tumour)]; + refs = map getref normals ++ optionals (tumour != null) [ (getref tumour) ]; ref = head refs; renameAndIndex = f: stage { diff --git a/tools/manta.nix b/tools/manta.nix index b34753f..dba8669 100644 --- a/tools/manta.nix +++ b/tools/manta.nix @@ -4,7 +4,7 @@ with bionix; { /* Call structural variants - Type: { ... } -> { normals :: [bam], tumour :: bam } -> manta + Type: { ... } -> { normals :: [bam], tumour :: bam } -> manta */ call = callBionixE ./manta-call.nix; } diff --git a/tools/minimap2-align.nix b/tools/minimap2-align.nix index a6de0b7..18bca35 100644 --- a/tools/minimap2-align.nix +++ b/tools/minimap2-align.nix @@ -18,7 +18,8 @@ let fa = f: matchFiletype "minimap2-ref" { fa = _: f; } f; fq = f: matchFiletype "minimap2-input" { fq = _: f; gz = matchFiletype' "minimap2-input" { fq = _: f; }; } f; -in stage { +in +stage { name = "minimap2-align"; buildInputs = with pkgs; [ minimap2 bc ] ++ optional bamOutput samtools; buildCommand = '' @@ -32,6 +33,6 @@ in stage { ${optionalString bamOutput "| samtools view -b"} \ > $out ''; - passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.none {};}; + passthru.filetype = if bamOutput then filetype.bam { inherit ref; sorting = sort.none { }; } else filetype.sam { inherit ref; sorting = sort.none { }; }; passthru.multicore = true; } diff --git a/tools/minimap2.nix b/tools/minimap2.nix index e1d6af5..59439ef 100644 --- a/tools/minimap2.nix +++ b/tools/minimap2.nix @@ -4,7 +4,7 @@ with bionix; rec { /* Align read against a reference - * Type: align :: {ref :: fasta, bamOutput :: bool, preset :: string, ...} -> {input1, input2} -> bam/sam + * Type: align :: {ref :: fasta, bamOutput :: bool, preset :: string, ...} -> {input1, input2} -> bam/sam */ align = callBionixE ./minimap2-align.nix; } diff --git a/tools/mosdepth-depth.nix b/tools/mosdepth-depth.nix index 2dbfc8d..2a72989 100644 --- a/tools/mosdepth-depth.nix +++ b/tools/mosdepth-depth.nix @@ -1,5 +1,5 @@ { bionix -, indexAttrs ? {} +, indexAttrs ? { } , flags ? null }: diff --git a/tools/mosdepth-plot.nix b/tools/mosdepth-plot.nix index f98ae23..87eb67f 100644 --- a/tools/mosdepth-plot.nix +++ b/tools/mosdepth-plot.nix @@ -7,7 +7,8 @@ with bionix; with lib; { inputs -, names ? []}: +, names ? [ ] +}: stage { name = "mosdepth-plot"; diff --git a/tools/mosdepth.nix b/tools/mosdepth.nix index cbebdfb..91c30e5 100644 --- a/tools/mosdepth.nix +++ b/tools/mosdepth.nix @@ -4,12 +4,12 @@ with bionix; { /* Compute coverage of a reference given an alignemnt - Type: { ... } -> bam -> mosdepth + Type: { ... } -> bam -> mosdepth */ depth = callBionixE ./mosdepth-depth.nix; /* Plot sample coverages. Names are optional. - Type: { ... } -> { inputs :: [mosdepth], names :: [string] } -> html + Type: { ... } -> { inputs :: [mosdepth], names :: [string] } -> html */ plot = callBionixE ./mosdepth-plot.nix; } diff --git a/tools/mutect-app.nix b/tools/mutect-app.nix index ca78d34..ad1515c 100644 --- a/tools/mutect-app.nix +++ b/tools/mutect-app.nix @@ -1,16 +1,19 @@ -{stdenv, fetchurl, makeWrapper, unzip, fetchFromGitHub}: +{ stdenv, fetchurl, makeWrapper, unzip, fetchFromGitHub }: let - oldnix = import (fetchFromGitHub { - owner = "NixOS"; - repo = "nixpkgs"; - rev = "83a893c38a83877588e3ca7ccfeabaa973c30acd"; - sha256 = "0q7214hag7h95irvhkdb648m09b9jspb0raw1qjrx7y4grzb165h"; - }) {}; + oldnix = import + (fetchFromGitHub { + owner = "NixOS"; + repo = "nixpkgs"; + rev = "83a893c38a83877588e3ca7ccfeabaa973c30acd"; + sha256 = "0q7214hag7h95irvhkdb648m09b9jspb0raw1qjrx7y4grzb165h"; + }) + { }; jre = oldnix.openjdk7; -in stdenv.mkDerivation rec { +in +stdenv.mkDerivation rec { name = "mutect-${version}"; version = "1.1.5"; diff --git a/tools/mutect-call.nix b/tools/mutect-call.nix index e98ba4b..52c1e1d 100644 --- a/tools/mutect-call.nix +++ b/tools/mutect-call.nix @@ -1,21 +1,22 @@ -{bionix +{ bionix , cosmic -, dbsnp}: +, dbsnp +}: with bionix; with lib; let inherit (types) matchFiletype; - getVCFref = matchFiletype "mutect-call" {vcf = {ref}: ref;}; - getBAMref = matchFiletype "mutect-call" {bam = {ref, ...}: ref;}; + getVCFref = matchFiletype "mutect-call" { vcf = { ref }: ref; }; + getBAMref = matchFiletype "mutect-call" { bam = { ref, ... }: ref; }; refs = map getVCFref [ cosmic dbsnp ]; ref = head refs; in assert (length (unique refs) == 1); -{normal, tumour}: +{ normal, tumour }: assert (ref == getBAMref normal && ref == getBAMref tumour); diff --git a/tools/mutect.nix b/tools/mutect.nix index addea77..8f40dee 100644 --- a/tools/mutect.nix +++ b/tools/mutect.nix @@ -1,12 +1,12 @@ -{bionix}: +{ bionix }: with bionix; { - app = pkgs.callPackage ./mutect-app.nix {}; + app = pkgs.callPackage ./mutect-app.nix { }; /* Call somatic variants with mutect - Type: { cosmic, dbsnp, ... } -> { normal :: bam, tumour :: bam } -> vcf + Type: { cosmic, dbsnp, ... } -> { normal :: bam, tumour :: bam } -> vcf */ call = callBionixE ./mutect-call.nix; } diff --git a/tools/octopus-call.nix b/tools/octopus-call.nix index 80092e0..81619b6 100644 --- a/tools/octopus-call.nix +++ b/tools/octopus-call.nix @@ -18,7 +18,7 @@ with types; inputs: let - getref = f: matchFiletype "octopus-call" { bam = { ref, ... }: ref; cram = { ref, ... }: ref; } f; + getref = matchFiletype "octopus-call" { bam = { ref, ... }: ref; cram = { ref, ... }: ref; }; refs = map getref inputs; ref = head refs; @@ -63,6 +63,6 @@ stage { ${optionalString (targets != null) (handleTarget targets)} \ ${flags} ''; - passthru.filetype = filetype.vcf { ref = ref; }; + passthru.filetype = filetype.vcf { inherit ref; }; passthru.multicore = true; } diff --git a/tools/octopus-callSomatic.nix b/tools/octopus-callSomatic.nix index 401eeb8..df41550 100644 --- a/tools/octopus-callSomatic.nix +++ b/tools/octopus-callSomatic.nix @@ -31,7 +31,7 @@ let ''; inputs = [ normal ] ++ tumours; - getref = f: matchFiletype "octopus-callSomatic" { bam = { ref, ... }: ref; cram = { ref, ... }: ref; } f; + getref = matchFiletype "octopus-callSomatic" { bam = { ref, ... }: ref; cram = { ref, ... }: ref; }; refs = map getref inputs; ref = head refs; @@ -77,6 +77,6 @@ stage { -N $normal \ ${flags} ''; - passthru.filetype = filetype.vcf { ref = ref; }; + passthru.filetype = filetype.vcf { inherit ref; }; passthru.multicore = true; } diff --git a/tools/octopus.nix b/tools/octopus.nix index 858d133..7c21758 100644 --- a/tools/octopus.nix +++ b/tools/octopus.nix @@ -4,12 +4,12 @@ with bionix; { /* Call variants for a population - Type: { fast :: bool, very-fast :: bool, max-genotypes :: int, targets :: FilePath + [string], ... } -> [bam] -> vcf + Type: { fast :: bool, very-fast :: bool, max-genotypes :: int, targets :: FilePath + [string], ... } -> [bam] -> vcf */ call = callBionixE ./octopus-call.nix; - + /* Call somatic variants - Type: { fast :: bool, very-fast :: bool, max-genotypes :: int, targets :: FilePath + [string], ... } -> { normal :: bam, tumours :: [bam] } -> vcf + Type: { fast :: bool, very-fast :: bool, max-genotypes :: int, targets :: FilePath + [string], ... } -> { normal :: bam, tumours :: [bam] } -> vcf */ callSomatic = callBionixE ./octopus-callSomatic.nix; } diff --git a/tools/picard-markDuplicates.nix b/tools/picard-markDuplicates.nix index 957a05e..e4b6369 100644 --- a/tools/picard-markDuplicates.nix +++ b/tools/picard-markDuplicates.nix @@ -1,8 +1,8 @@ { bionix , flags ? null -} : +}: -inputBam : +inputBam: with bionix; with lib; @@ -10,24 +10,24 @@ with types; assert (matchFiletype "picard-markDuplicates" { bam = _: true; } inputBam); assert (matchFileSorting - "picard-markDuplicates" - { coord = _: true; name = _: true; } - inputBam); + "picard-markDuplicates" + { coord = _: true; name = _: true; } + inputBam); # Note that picard markDuplicates has different behaviour depending on whether the input # is name-sorted or coordinate-sorted. stage { - name = "picard-markDuplicates"; - buildInputs = with pkgs; - [ picard-tools ]; - outputs = [ "out" "metrics" ]; - buildCommand = '' - picard MarkDuplicates \ - I=${inputBam} \ - O=$out \ - M=$metrics \ - ${optionalString (flags != null) flags} - ''; - passthru.filetype = inputBam.filetype; + name = "picard-markDuplicates"; + buildInputs = with pkgs; + [ picard-tools ]; + outputs = [ "out" "metrics" ]; + buildCommand = '' + picard MarkDuplicates \ + I=${inputBam} \ + O=$out \ + M=$metrics \ + ${optionalString (flags != null) flags} + ''; + passthru.filetype = inputBam.filetype; } diff --git a/tools/picard.nix b/tools/picard.nix index 4246afe..3cb61a1 100644 --- a/tools/picard.nix +++ b/tools/picard.nix @@ -3,8 +3,8 @@ with bionix; { - /* Mark duplicates + /* Mark duplicates Type: { ... } -> bam -> bam - */ - markDuplicates = callBionixE ./picard-markDuplicates.nix; -}
\ No newline at end of file + */ + markDuplicates = callBionixE ./picard-markDuplicates.nix; +} diff --git a/tools/platypus-callVariants.nix b/tools/platypus-callVariants.nix index bb0faf1..59a224b 100644 --- a/tools/platypus-callVariants.nix +++ b/tools/platypus-callVariants.nix @@ -1,6 +1,6 @@ { bionix -, indexAttrs ? {} -, bamIndexAttrs ? {} +, indexAttrs ? { } +, bamIndexAttrs ? { } , flags ? null }: @@ -13,7 +13,7 @@ with types; let filename = path: last (splitString "/" path); - getref = f: matchFiletype "platypus-callVariants" { bam = {ref, ...}: ref; } f; + getref = matchFiletype "platypus-callVariants" { bam = { ref, ... }: ref; }; refs = map getref inputs; ref = head refs; in @@ -38,6 +38,6 @@ stage { # Remove timestamps from output sed -i '/^##fileDate/d' $out ''; - passthru.filetype = filetype.vcf {ref = ref;}; + passthru.filetype = filetype.vcf { inherit ref; }; passthru.multicore = true; } diff --git a/tools/platypus.nix b/tools/platypus.nix index c93a2d8..29e7e08 100644 --- a/tools/platypus.nix +++ b/tools/platypus.nix @@ -4,7 +4,7 @@ with bionix; { /* Call variants - Type: { ... } -> [bam] -> vcf + Type: { ... } -> [bam] -> vcf */ call = callBionixE ./platypus-callVariants.nix; } diff --git a/tools/sambamba.nix b/tools/sambamba.nix index 3c1d1fd..77289fe 100644 --- a/tools/sambamba.nix +++ b/tools/sambamba.nix @@ -5,34 +5,35 @@ with bionix; let gen = callBionixE ./sambamba-generic.nix; -in { +in +{ /* Sort aligned reads - Type: { nameSort :: bool, ... } -> bam -> bam + Type: { nameSort :: bool, ... } -> bam -> bam */ sort = callBionixE ./sambamba-sort.nix; /* Build an index - Type: { ... } -> bam -> index + Type: { ... } -> bam -> index */ - index = def gen {tool = "index"; }; + index = def gen { tool = "index"; }; /* Merge bam files - Type: { ... } -> [bam] -> bam + Type: { ... } -> [bam] -> bam */ - merge = def gen {tool = "merge"; }; + merge = def gen { tool = "merge"; }; /* Slice a region out of a bam file - Type: { region, ... } -> bam -> bam + Type: { region, ... } -> bam -> bam */ - slice = def gen {tool = "slice"; }; + slice = def gen { tool = "slice"; }; /* Compute flag statistics - Type: { ... } -> bam -> flagstat + Type: { ... } -> bam -> flagstat */ - flagstat = def gen {tool = "flagstat"; }; + flagstat = def gen { tool = "flagstat"; }; /* Mark duplicates - Type: { ... } -> bam -> bam + Type: { ... } -> bam -> bam */ - markdup = def gen {tool = "markdup"; }; + markdup = def gen { tool = "markdup"; }; } diff --git a/tools/samtools-merge.nix b/tools/samtools-merge.nix index 792ee52..ff8ea00 100644 --- a/tools/samtools-merge.nix +++ b/tools/samtools-merge.nix @@ -10,8 +10,8 @@ with lib; let inherit (bionix.types) matchFiletype matchSorting; - inputIsHomogenous = length (unique (map (matchFiletype "samtools-merge" {bam = x: x // {sorting = matchSorting "samtools-merge" {coord = _: "coord"; name = _: "name"; } x;};}) inputs)) == 1; - nameSorted = matchFiletype "samtools-merge" { bam = x: matchSorting "samtools-merge" {coord = _: false; name = _: true;} x;} (lib.head inputs); + inputIsHomogenous = length (unique (map (matchFiletype "samtools-merge" { bam = x: x // { sorting = matchSorting "samtools-merge" { coord = _: "coord"; name = _: "name"; } x; }; }) inputs)) == 1; + nameSorted = matchFiletype "samtools-merge" { bam = matchSorting "samtools-merge" { coord = _: false; name = _: true; }; } (lib.head inputs); in assert inputIsHomogenous; diff --git a/tools/samtools-sort.nix b/tools/samtools-sort.nix index e3fbff8..30439f1 100644 --- a/tools/samtools-sort.nix +++ b/tools/samtools-sort.nix @@ -17,9 +17,10 @@ assert (matchFiletype "samtools-sort" { bam = _: true; sam = _: true; cram = _: let outfmtR = if outfmt != null then outfmt input else input.filetype; - outFmtFlags = matchFiletype "samtools-sort-outfmt" { bam = _: "-O BAM"; sam = _: "-O SAM"; cram = ref: "-O CRAM -T ${ref}"; } {filetype = outfmtR;}; - alreadySorted = matchFileSorting "samtools-sort" { name = _: nameSort; coord = _: !nameSort; none = _: false;} input; -in stage { + outFmtFlags = matchFiletype "samtools-sort-outfmt" { bam = _: "-O BAM"; sam = _: "-O SAM"; cram = ref: "-O CRAM -T ${ref}"; } { filetype = outfmtR; }; + alreadySorted = matchFileSorting "samtools-sort" { name = _: nameSort; coord = _: !nameSort; none = _: false; } input; +in +stage { name = "samtools-sort"; buildInputs = with pkgs; [ samtools ]; buildCommand = diff --git a/tools/samtools-view.nix b/tools/samtools-view.nix index 35a3fcb..4d2bdc7 100644 --- a/tools/samtools-view.nix +++ b/tools/samtools-view.nix @@ -15,8 +15,9 @@ assert (matchFiletype "samtools-view" { bam = _: true; sam = _: true; cram = _: let outfmtR = if outfmt != null then (if builtins.typeOf outfmt == "string" then { "bam" = toBam; "cram" = toCram; "sam" = toSam; }."${outfmt}" else outfmt) input else input.filetype; fa = ref: matchFiletype "samtools-view-ref" { fa = _: ref; } ref; - outfmtFlags = matchFiletype "samtools-view-outfmt" { bam = _: "-O BAM"; sam = _: "-O SAM"; cram = x: "-O CRAM -T ${fa x.ref}"; } {filetype = outfmtR;}; -in stage { + outfmtFlags = matchFiletype "samtools-view-outfmt" { bam = _: "-O BAM"; sam = _: "-O SAM"; cram = x: "-O CRAM -T ${fa x.ref}"; } { filetype = outfmtR; }; +in +stage { name = "samtools-view"; buildInputs = with pkgs; [ samtools ]; buildCommand = '' diff --git a/tools/snap-align.nix b/tools/snap-align.nix index 279cf20..f0bcb21 100644 --- a/tools/snap-align.nix +++ b/tools/snap-align.nix @@ -2,7 +2,7 @@ , ref , bamOutput ? true , flags ? null -, indexAttrs ? {} +, indexAttrs ? { } }: { input1 @@ -18,7 +18,8 @@ let fa = f: matchFiletype "snap-ref" { fa = _: f; } f; fq = f: matchFiletype "snap-input" { fq = _: f; gz = matchFiletype' "snap-input" { fq = _: f; }; } f; -in stage { +in +stage { name = "snap-align"; buildInputs = with pkgs; [ bionix.snap.app bc ] ++ optional bamOutput samtools; buildCommand = '' @@ -32,6 +33,6 @@ in stage { ${optionalString (flags != null) flags} \ | samtools sort -n > $out ''; - passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};}; + passthru.filetype = if bamOutput then filetype.bam { inherit ref; sorting = sort.none { }; } else filetype.sam { inherit ref; sorting = sort.name { }; }; passthru.multicore = true; } diff --git a/tools/snap-app.nix b/tools/snap-app.nix index c845d1f..0e74a20 100644 --- a/tools/snap-app.nix +++ b/tools/snap-app.nix @@ -1,4 +1,4 @@ -{stdenv, fetchFromGitHub, zlib}: +{ stdenv, fetchFromGitHub, zlib }: stdenv.mkDerivation rec { name = "snap-git"; diff --git a/tools/snap.nix b/tools/snap.nix index 851c51a..51df5dc 100644 --- a/tools/snap.nix +++ b/tools/snap.nix @@ -3,15 +3,15 @@ with bionix; rec { - app = pkgs.callPackage ./snap-app.nix {}; + app = pkgs.callPackage ./snap-app.nix { }; /* Align reads against a reference - Type: snap :: {ref :: fasta, bamOutput :: bool, ...} -> {input1, input2} -> bam/sam + Type: snap :: {ref :: fasta, bamOutput :: bool, ...} -> {input1, input2} -> bam/sam */ align = callBionixE ./snap-align.nix; /* Creates an reference index for SNAP - Type: index :: {...} -> fasta -> SNAP index + Type: index :: {...} -> fasta -> SNAP index */ index = callBionixE ./snap-index.nix; } diff --git a/tools/snpeff-annotate.nix b/tools/snpeff-annotate.nix index 317d303..9324f47 100644 --- a/tools/snpeff-annotate.nix +++ b/tools/snpeff-annotate.nix @@ -1,7 +1,8 @@ -{bionix -,db -,heapSize ? "31g" -,flags ? ""}: +{ bionix +, db +, heapSize ? "31g" +, flags ? "" +}: input: diff --git a/tools/snpeff-dbnsfp.nix b/tools/snpeff-dbnsfp.nix index 518c121..003f7f8 100644 --- a/tools/snpeff-dbnsfp.nix +++ b/tools/snpeff-dbnsfp.nix @@ -1,7 +1,8 @@ -{bionix -,dbnsfp -,heapSize ? "31g" -,flags ? ""}: +{ bionix +, dbnsfp +, heapSize ? "31g" +, flags ? "" +}: input: diff --git a/tools/snpeff.nix b/tools/snpeff.nix index fb3f4a9..45a0edc 100644 --- a/tools/snpeff.nix +++ b/tools/snpeff.nix @@ -1,15 +1,15 @@ -{bionix}: +{ bionix }: with bionix; { /* Annotate variants with SNPEff database. Some annotation DBs available in bionix.ref (e.g., bionix.ref.grch38.snpeff.db). - Type: { db, ... } -> vcf -> vcf + Type: { db, ... } -> vcf -> vcf */ annotate = callBionixE ./snpeff-annotate.nix; - + /* Annotate variants with dbNSFP database. Some dbNSFP annotation DBs available in bionix.ref (e.g., bionix.ref.grch38.snpeff.dbNSFP). - Type: { dbnsfp, ... } -> vcf -> vcf + Type: { dbnsfp, ... } -> vcf -> vcf */ dbnsfp = callBionixE ./snpeff-dbnsfp.nix; } diff --git a/tools/snver-call.nix b/tools/snver-call.nix index 72562c7..2ad4b72 100644 --- a/tools/snver-call.nix +++ b/tools/snver-call.nix @@ -11,7 +11,7 @@ with types; let config = pkgs.writeText "pool.txt" (concatMapStringsSep "\n" (x: "${x}\t${toString ploidy}\t1") (attrValues inputs)); - getref = f: matchFiletype "SNVer-call" { bam = {ref, ...}: ref; } f; + getref = matchFiletype "SNVer-call" { bam = { ref, ... }: ref; }; refs = mapAttrsToList (_: getref) inputs; ref = head refs; in diff --git a/tools/snver.nix b/tools/snver.nix index 6f95e1b..c5799ec 100644 --- a/tools/snver.nix +++ b/tools/snver.nix @@ -4,10 +4,10 @@ with bionix; with pkgs; { - app = callPackage ./snver-app.nix {}; + app = callPackage ./snver-app.nix { }; /* Call variants - Type: { ploidy, ... } -> [bam] -> vcf + Type: { ploidy, ... } -> [bam] -> vcf */ call = callBionix ./snver-call.nix; } diff --git a/tools/star-align.nix b/tools/star-align.nix index 44dd7f2..670c6bf 100644 --- a/tools/star-align.nix +++ b/tools/star-align.nix @@ -2,7 +2,7 @@ , ref , bamOutput ? true , flags ? null -, indexAttrs ? {} +, indexAttrs ? { } }: { input1 @@ -18,7 +18,8 @@ let fa = f: matchFiletype "star-ref" { fa = _: f; } f; fq = f: matchFiletype "star-input" { fq = _: f; gz = matchFiletype' "star-input" { fq = _: "<(gunzip < ${f})"; }; } f; -in stage { +in +stage { name = "star-align"; buildInputs = with pkgs; [ star bc samtools ]; buildCommand = '' @@ -33,6 +34,6 @@ in stage { --readFilesIn ${fq input1} ${optionalString (input2 != null) (fq input2)} ${if bamOutput then "samtools view -b Aligned.out.sam > $out" else "cp Aligned.out.sam $out"} ''; - passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};}; + passthru.filetype = if bamOutput then filetype.bam { inherit ref; sorting = sort.none { }; } else filetype.sam { inherit ref; sorting = sort.name { }; }; passthru.multicore = true; } diff --git a/tools/star-index.nix b/tools/star-index.nix index 4bc8399..9b7c819 100644 --- a/tools/star-index.nix +++ b/tools/star-index.nix @@ -1,8 +1,8 @@ { bionix , gtf , flags ? null -, extractSpliceSitesAttrs ? {} -, extractExonsAttrs ? {} +, extractSpliceSitesAttrs ? { } +, extractExonsAttrs ? { } , overhang ? 100 }: diff --git a/tools/star.nix b/tools/star.nix index 744c4e0..1209d4d 100644 --- a/tools/star.nix +++ b/tools/star.nix @@ -4,7 +4,7 @@ with bionix; { /* Align RNA against a reference genome - Type: { ref, ... } -> { input1, input2 } -> bam + Type: { ref, ... } -> { input1, input2 } -> bam */ align = callBionixE ./star-align.nix; index = callBionixE ./star-index.nix; diff --git a/tools/strelka-call.nix b/tools/strelka-call.nix index 74dfd0b..6a84b24 100644 --- a/tools/strelka-call.nix +++ b/tools/strelka-call.nix @@ -1,6 +1,6 @@ { bionix -, indexAttrs ? {} -, bamIndexAttrs ? {} +, indexAttrs ? { } +, bamIndexAttrs ? { } , flags ? null }: @@ -12,7 +12,7 @@ with types; let filename = path: last (splitString "/" path); - getref = f: matchFiletype "strelka-call" { bam = x: x.ref; } f; + getref = matchFiletype "strelka-call" { bam = x: x.ref; }; refs = map getref inputs; ref = head refs; @@ -21,38 +21,39 @@ in assert (length (unique refs) == 1); let -out = stage { - name = "strelka-call"; - buildInputs = with pkgs; [ strelka ]; - buildCommand = '' - ln -s ${ref} ref.fa - ln -s ${bionix.samtools.faidx indexAttrs ref} ref.fa.fai - ${concatMapStringsSep "\n" (p: "ln -s ${p} ${filename p}.bam") inputs} - ${concatMapStringsSep "\n" (p: "ln -s ${bionix.samtools.index bamIndexAttrs p} ${filename p}.bai") inputs} + out = stage { + name = "strelka-call"; + buildInputs = with pkgs; [ strelka ]; + buildCommand = '' + ln -s ${ref} ref.fa + ln -s ${bionix.samtools.faidx indexAttrs ref} ref.fa.fai + ${concatMapStringsSep "\n" (p: "ln -s ${p} ${filename p}.bam") inputs} + ${concatMapStringsSep "\n" (p: "ln -s ${bionix.samtools.index bamIndexAttrs p} ${filename p}.bai") inputs} - configureStrelkaGermlineWorkflow.py \ - ${concatMapStringsSep " " (i: "--bam ${filename i}.bam") inputs} \ - --ref ref.fa \ - --runDir $TMPDIR + configureStrelkaGermlineWorkflow.py \ + ${concatMapStringsSep " " (i: "--bam ${filename i}.bam") inputs} \ + --ref ref.fa \ + --runDir $TMPDIR - ./runWorkflow.py \ - -m local \ - -j $NIX_BUILD_CORES 2>&1 + ./runWorkflow.py \ + -m local \ + -j $NIX_BUILD_CORES 2>&1 - # Strelka writes runtime stats and timestamps; - # both have to be stripped to provide determinism - cd results/variants - rm *.tbi genome.vcf.gz - for f in *.vcf.gz; do - gunzip $f - g=$(basename $f .gz) - sed -i '/^##fileDate/d' $g - sed -i '/^##startTime/d' $g - sed -i '/^##cmd/d' $g - done - mv variants.vcf $out - ''; - passthru.multicore = true; - passthru.filetype = types.filetype.vcf {ref = ref;}; -}; -in out + # Strelka writes runtime stats and timestamps; + # both have to be stripped to provide determinism + cd results/variants + rm *.tbi genome.vcf.gz + for f in *.vcf.gz; do + gunzip $f + g=$(basename $f .gz) + sed -i '/^##fileDate/d' $g + sed -i '/^##startTime/d' $g + sed -i '/^##cmd/d' $g + done + mv variants.vcf $out + ''; + passthru.multicore = true; + passthru.filetype = types.filetype.vcf { inherit ref; }; + }; +in +out diff --git a/tools/strelka-callSomatic.nix b/tools/strelka-callSomatic.nix index b409344..6a8f252 100644 --- a/tools/strelka-callSomatic.nix +++ b/tools/strelka-callSomatic.nix @@ -1,10 +1,10 @@ { bionix -, indexAttrs ? {} -, bamIndexAttrs ? {} +, indexAttrs ? { } +, bamIndexAttrs ? { } , flags ? null }: -{normal, tumour}: +{ normal, tumour }: with bionix; with lib; @@ -12,7 +12,7 @@ with types; let filename = path: last (splitString "/" path); - getref = f: matchFiletype "strelka-callSomatic" { bam = x: x.ref; } f; + getref = matchFiletype "strelka-callSomatic" { bam = x: x.ref; }; inputs = [ normal tumour ]; refs = map getref inputs; ref = head refs; @@ -23,44 +23,45 @@ assert (length (unique refs) == 1); let -out = stage { - name = "strelka-callSomatic"; - buildInputs = with pkgs; [ strelka ]; - outputs = [ "out" "indels" "snvs" ]; - buildCommand = '' - ln -s ${ref} ref.fa - ln -s ${bionix.samtools.faidx indexAttrs ref} ref.fa.fai - ${concatMapStringsSep "\n" (p: "ln -s ${p} ${filename p}.bam") inputs} - ${concatMapStringsSep "\n" (p: "ln -s ${bionix.samtools.index bamIndexAttrs p} ${filename p}.bai") inputs} + out = stage { + name = "strelka-callSomatic"; + buildInputs = with pkgs; [ strelka ]; + outputs = [ "out" "indels" "snvs" ]; + buildCommand = '' + ln -s ${ref} ref.fa + ln -s ${bionix.samtools.faidx indexAttrs ref} ref.fa.fai + ${concatMapStringsSep "\n" (p: "ln -s ${p} ${filename p}.bam") inputs} + ${concatMapStringsSep "\n" (p: "ln -s ${bionix.samtools.index bamIndexAttrs p} ${filename p}.bai") inputs} - configureStrelkaSomaticWorkflow.py \ - --normalBam ${filename normal}.bam \ - --tumourBam ${filename tumour}.bam \ - --ref ref.fa \ - --runDir $TMPDIR + configureStrelkaSomaticWorkflow.py \ + --normalBam ${filename normal}.bam \ + --tumourBam ${filename tumour}.bam \ + --ref ref.fa \ + --runDir $TMPDIR - ./runWorkflow.py \ - -m local \ - -j $NIX_BUILD_CORES + ./runWorkflow.py \ + -m local \ + -j $NIX_BUILD_CORES - # Strelka writes runtime stats and timestamps; - # both have to be stripped to provide determinism - cd results/variants - rm *.tbi - for f in *.vcf.gz; do - gunzip $f - g=$(basename $f .gz) - sed -i '/^##fileDate/d' $g - sed -i '/^##startTime/d' $g - sed -i '/^##cmd/d' $g - done - mv somatic.indels.vcf $indels - mv somatic.snvs.vcf $snvs + # Strelka writes runtime stats and timestamps; + # both have to be stripped to provide determinism + cd results/variants + rm *.tbi + for f in *.vcf.gz; do + gunzip $f + g=$(basename $f .gz) + sed -i '/^##fileDate/d' $g + sed -i '/^##startTime/d' $g + sed -i '/^##cmd/d' $g + done + mv somatic.indels.vcf $indels + mv somatic.snvs.vcf $snvs - ln -s $snvs $out - ''; - passthru.multicore = true; - passthru.filetype = types.filetype.vcf {ref = ref;}; -}; + ln -s $snvs $out + ''; + passthru.multicore = true; + passthru.filetype = types.filetype.vcf { inherit ref; }; + }; -in out +in +out diff --git a/tools/strelka.nix b/tools/strelka.nix index 16f4a2b..c71f07b 100644 --- a/tools/strelka.nix +++ b/tools/strelka.nix @@ -5,11 +5,11 @@ with types; { /* Calls somatic variants - Type: callSomatic :: {...} -> {tumour, normal} -> somatic results + Type: callSomatic :: {...} -> {tumour, normal} -> somatic results */ callSomatic = callBionixE ./strelka-callSomatic.nix; /* Calls variants - Type: call :: {...} -> [input] -> results + Type: call :: {...} -> [input] -> results */ call = callBionixE ./strelka-call.nix; } diff --git a/tools/whisper-align.nix b/tools/whisper-align.nix index 1731bfe..19eef0d 100644 --- a/tools/whisper-align.nix +++ b/tools/whisper-align.nix @@ -2,7 +2,7 @@ , ref , bamOutput ? true , flags ? null -, indexAttrs ? {} +, indexAttrs ? { } }: { input1 @@ -18,7 +18,8 @@ let fa = f: matchFiletype "whisper-ref" { fa = _: f; } f; fq = f: matchFiletype "whisper-input" { fq = _: f; gz = matchFiletype' "whisper-input" { fq = _: f; }; } f; -in stage { +in +stage { name = "whisper-mem"; buildInputs = with pkgs; [ whisper ]; buildCommand = '' @@ -34,6 +35,6 @@ in stage { ${optionalString (input2 != null) (fq input2)} \ > $out ''; - passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};}; + passthru.filetype = if bamOutput then filetype.bam { inherit ref; sorting = sort.none { }; } else filetype.sam { inherit ref; sorting = sort.name { }; }; passthru.multicore = true; } diff --git a/tools/whisper.nix b/tools/whisper.nix index ff56f8a..8b89cb2 100644 --- a/tools/whisper.nix +++ b/tools/whisper.nix @@ -6,7 +6,7 @@ rec { index = callBionixE ./whisper-index.nix; /* Align reads against a reference - Type: { ref, ... } -> { input1, input2 } -> bam + Type: { ref, ... } -> { input1, input2 } -> bam */ align = callBionixE ./whisper-align.nix; } diff --git a/tools/xenomapper-allocate.nix b/tools/xenomapper-allocate.nix index 3c0fc08..adac472 100644 --- a/tools/xenomapper-allocate.nix +++ b/tools/xenomapper-allocate.nix @@ -2,7 +2,7 @@ , flags ? null }: -{primary, secondary}: +{ primary, secondary }: with bionix; with lib; @@ -10,8 +10,8 @@ with types; let - isSortedBam = matchFiletype "xenomapper-allocate" {bam = matchSorting "xenomapper-allocate" { coord = _: false; name = _: true; none = _: false; }; }; - outs = [ "primary_specific" "primary_multi" "secondary_specific" "secondary_multi" "unassigned" "unresolved"]; + isSortedBam = matchFiletype "xenomapper-allocate" { bam = matchSorting "xenomapper-allocate" { coord = _: false; name = _: true; none = _: false; }; }; + outs = [ "primary_specific" "primary_multi" "secondary_specific" "secondary_multi" "unassigned" "unresolved" ]; in diff --git a/tools/xenomapper.nix b/tools/xenomapper.nix index d22f84b..5ac903e 100644 --- a/tools/xenomapper.nix +++ b/tools/xenomapper.nix @@ -1,10 +1,10 @@ -{bionix}: +{ bionix }: with bionix; { /* Split aligned reads based on primary and secondary alignments - Type: { ... } -> { primary :: bam, secondary :: bam } -> { primary_specific :: bam, primary_multi :: bam, secondary_specific :: bam, secondary_multi :: bam, unassigned :: bam, unresolved :: bam } + Type: { ... } -> { primary :: bam, secondary :: bam } -> { primary_specific :: bam, primary_multi :: bam, secondary_specific :: bam, secondary_multi :: bam, unassigned :: bam, unresolved :: bam } */ allocate = callBionixE ./xenomapper-allocate.nix; -}
\ No newline at end of file +} |