Merge branch 'master' of https://github.com/PapenfussLab/bionix

18 files changed, 170 insertions, 6 deletions

diff --git a/default.nix b/default.nix
index 5218a99..f835fa0 100644
--- a/default.nix
+++ b/default.nix
@@ -15,6 +15,7 @@ let
     fastqc = callBionix ./tools/fastqc.nix {};
     gridss = callBionix ./tools/gridss.nix {};
     infercnv = callBionix ./tools/infercnv.nix {};
+    kallisto = callBionix ./tools/kallisto.nix {};
     mosdepth = callBionix ./tools/mosdepth.nix {};
     mutect = callBionix ./tools/mutect.nix {};
     platypus = callBionix ./tools/platypus.nix {};
diff --git a/lib/references.nix b/lib/references.nix
index 498cfd9..bdeefa6 100755
--- a/lib/references.nix
+++ b/lib/references.nix
@@ -46,6 +46,28 @@ rec {
       '';
       passthru.filetype = filetype.vcf { ref = seq; };
     };
+    ensembl = {
+      cdna = stdenvNoCC.mkDerivation rec {
+        name = "ensembl-grch38-cdna-${version}";
+        version = "94";
+        src = fetchurl {
+          url = "ftp://ftp.ensembl.org/pub/release-${version}/fasta/homo_sapiens/cdna/Homo_sapiens.GRCh38.cdna.all.fa.gz";
+          sha256 = "1fc5d6p2wlwsm49wnmxmm3byjx5jvr6z9fpzrq7v7fpb086adl0h";
+        };
+        buildCommand = "gunzip < $src > $out";
+        passthru.filetype = filetype.fa {};
+      };
+      ncrna = stdenvNoCC.mkDerivation rec {
+        name = "ensembl-grch38-ncrna-${version}";
+        version = "94";
+        src = fetchurl {
+          url = "ftp://ftp.ensembl.org/pub/release-${version}/fasta/homo_sapiens/ncrna/Homo_sapiens.GRCh38.ncrna.fa.gz";
+          sha256 = "1cpasykwriila52nqgvw6d3mjyh6d9qi613hvhn4h1dxkqzgnjff";
+        };
+        buildCommand = "gunzip < $src > $out";
+        passthru.filetype = filetype.fa {};
+      };
+    };
   };
 
   grcm38 = grcm38-p6;
diff --git a/lib/types.nix b/lib/types.nix
index 4dbc6ff..ef9b1b7 100644
--- a/lib/types.nix
+++ b/lib/types.nix
@@ -21,7 +21,8 @@ let
 
 in
 rec {
-  matchFiletype = sym: y: x: if x ? filetype then match x.filetype (defError (idft sym) y filetype) else abort "unknown filetype for ${sym}";
+  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 = {};
diff --git a/tools/bwa-mem.nix b/tools/bwa-mem.nix
index 3420294..2b3bc64 100644
--- a/tools/bwa-mem.nix
+++ b/tools/bwa-mem.nix
@@ -17,7 +17,7 @@ with bionix.compression;
 
 let
   fa = f: matchFiletype "bwa-ref" { fa = _: f; } f;
-  fq = f: matchFiletype "bwa-input" { fq = _: f; } f;
+  fq = f: matchFiletype "bwa-input" { fq = _: f; gz = matchFiletype' "bwa-input" { fq = _: f; }; } f;
 
 in stdenv.mkDerivation {
   name = "bwa-mem";
diff --git a/tools/gridss-annotateVariants.nix b/tools/gridss-annotateVariants.nix
index 4f66c6c..33992c4 100644
--- a/tools/gridss-annotateVariants.nix
+++ b/tools/gridss-annotateVariants.nix
@@ -8,6 +8,7 @@
 , softClipsToSplitReadsAttrs ? {}
 , identifyVariantsAttrs ? {}
 , flags ? null
+, config ? null
 }:
 
 with nixpkgs;
@@ -64,6 +65,7 @@ stdenv.mkDerivation rec {
       INPUT_VCF=input.vcf \
       OUTPUT_VCF=out.vcf \
       WORKING_DIR=$TMPDIR/ \
+      ${optionalString config ("CONFIGURATION_FILE=" + gridssConfig config)} \
       TMP_DIR=$TMPDIR/
 
     mv out.vcf $out
diff --git a/tools/gridss-callVariants.nix b/tools/gridss-callVariants.nix
index 416eb7c..dc978fa 100644
--- a/tools/gridss-callVariants.nix
+++ b/tools/gridss-callVariants.nix
@@ -4,6 +4,7 @@
 , bwaIndexAttrs ? {}
 , faidxAttrs ? {}
 , flags ? null
+, config ? null
 }:
 
 with nixpkgs;
@@ -41,6 +42,7 @@ stdenv.mkDerivation rec {
       WORKER_THREADS=$NIX_BUILD_CORES \
 	    TMP_DIR=. \
 	    WORKING_DIR=. \
+      ${optionalString config ("CONFIGURATION_FILE=" + gridssConfig config)} \
 	    REFERENCE_SEQUENCE="ref.fa" \
       ${concatMapStringsSep " " (i: "INPUT=\"${i}\"") inputs} \
 	    OUTPUT="$out/gridss.vcf" \
diff --git a/tools/gridss-collectGridssMetrics.nix b/tools/gridss-collectGridssMetrics.nix
index bb8d2e9..86732b8 100644
--- a/tools/gridss-collectGridssMetrics.nix
+++ b/tools/gridss-collectGridssMetrics.nix
@@ -2,6 +2,7 @@
 , nixpkgs
 , thresholdCoverage ? 10000
 , flags ? null
+, config ? null
 }:
 
 with nixpkgs;
@@ -25,6 +26,7 @@ stdenv.mkDerivation rec {
     java -Xmx1G -cp ${bionix.gridss.jar} \
 			gridss.analysis.CollectGridssMetrics \
 			${optionalString sorted "ASSUME_SORTED=true"} \
+      ${optionalString config ("CONFIGURATION_FILE=" + gridssConfig config)} \
 			I=input.bam \
 			O=$out \
 			THRESHOLD_COVERAGE=${toString thresholdCoverage}
diff --git a/tools/gridss-collectMetrics.nix b/tools/gridss-collectMetrics.nix
index 4688808..c1909f1 100644
--- a/tools/gridss-collectMetrics.nix
+++ b/tools/gridss-collectMetrics.nix
@@ -2,6 +2,7 @@
 , nixpkgs
 , thresholdCoverage ? 10000
 , flags ? null
+, config ? null
 }:
 
 with nixpkgs;
@@ -22,6 +23,7 @@ stdenv.mkDerivation rec {
     mkdir $out
     java -Xmx1G -cp ${bionix.gridss.jar} \
 			gridss.analysis.CollectGridssMetrics \
+      ${optionalString config ("CONFIGURATION_FILE=" + gridssConfig config)} \
 			I=${input}\
 			O=$out/input \
       AS=true \
diff --git a/tools/gridss-computeSamTags.nix b/tools/gridss-computeSamTags.nix
index f75cea0..5b19a3a 100644
--- a/tools/gridss-computeSamTags.nix
+++ b/tools/gridss-computeSamTags.nix
@@ -4,6 +4,7 @@
 , bwaIndexAttrs ? {}
 , faidxAttrs ? {}
 , flags ? null
+, config ? null
 }:
 
 with nixpkgs;
@@ -34,6 +35,7 @@ stdenv.mkDerivation rec {
 			REFERENCE_SEQUENCE=ref.fa \
 			WORKING_DIR=$TMP_DIR \
 			TMP_DIR=$TMP_DIR \
+      ${optionalString config ("CONFIGURATION_FILE=" + gridssConfig config)} \
 			I=${input} \
 			O=$out \
       AS=true
diff --git a/tools/gridss-configFile.nix b/tools/gridss-configFile.nix
new file mode 100644
index 0000000..838c292
--- /dev/null
+++ b/tools/gridss-configFile.nix
@@ -0,0 +1,36 @@
+{bionix, nixpkgs}:
+
+with nixpkgs;
+
+let
+    attrsToGridssConfigString = attrsToGridssConfigStringPrepend "";
+
+    attrsToGridssConfigStringPrepend = prepend: attrs:
+        lib.concatStringsSep "\n" (
+            lib.attrValues (
+                lib.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 + "\".");
+
+    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");
+        attrs  = 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: (writeText
+        "gridss.properties.override"
+        (attrsToGridssConfigString configAttrs))
diff --git a/tools/gridss-extractSVReads.nix b/tools/gridss-extractSVReads.nix
index a10abf5..e3ed1cc 100644
--- a/tools/gridss-extractSVReads.nix
+++ b/tools/gridss-extractSVReads.nix
@@ -5,6 +5,7 @@
 , flags ? null
 , unmappedReads ? false
 , minClipLength ? 5
+, config ? null
 }:
 
 with nixpkgs;
@@ -35,6 +36,7 @@ stdenv.mkDerivation rec {
       METRICS_OUTPUT=$out/input.sv_metrics \
       INSERT_SIZE_METRICS=$out/input.insert_size_metrics \
       UNMAPPED_READS=${if unmappedReads then "true" else "false"} \
+      ${optionalString config ("CONFIGURATION_FILE=" + gridssConfig config)} \
       MIN_CLIP_LENGTH=${toString minClipLength}
   '';
 }
diff --git a/tools/gridss-identifyVariants.nix b/tools/gridss-identifyVariants.nix
index f44771b..16982e2 100644
--- a/tools/gridss-identifyVariants.nix
+++ b/tools/gridss-identifyVariants.nix
@@ -7,6 +7,7 @@
 , collectMetricsAttrs ? {}
 , softClipsToSplitReadsAttrs ? {}
 , flags ? null
+, config ? null
 }:
 
 with nixpkgs;
@@ -60,6 +61,7 @@ stdenv.mkDerivation rec {
       ${concatMapStringsSep " " (i: "INPUT='${i}'") inputs} \
       ASSEMBLY=${assembly} \
       OUTPUT_VCF=out.vcf \
+      ${optionalString config ("CONFIGURATION_FILE=" + gridssConfig config)} \
       WORKING_DIR=$TMPDIR/ \
       TMP_DIR=$TMPDIR/
 
diff --git a/tools/gridss-softClipsToSplitReads.nix b/tools/gridss-softClipsToSplitReads.nix
index cc40e52..b35d41d 100644
--- a/tools/gridss-softClipsToSplitReads.nix
+++ b/tools/gridss-softClipsToSplitReads.nix
@@ -4,6 +4,7 @@
 , faidxAttrs ? {}
 , alignerStreaming ? false
 , flags ? null
+, config ? null
 }:
 
 with nixpkgs;
@@ -33,6 +34,7 @@ stdenv.mkDerivation rec {
 			I=${input} \
 			O=$out \
       ${optionalString alignerStreaming "ALIGNER_STREAMING=true"} \
+      ${optionalString config ("CONFIGURATION_FILE=" + gridssConfig config)} \
 			WORKER_THREADS=$NIX_BUILD_CORES
     '';
   passthru.filetype =
diff --git a/tools/gridss.nix b/tools/gridss.nix
index 792d279..1e4e2d1 100644
--- a/tools/gridss.nix
+++ b/tools/gridss.nix
@@ -8,6 +8,7 @@ rec {
     url = "https://github.com/PapenfussLab/gridss/releases/download/v2.0.0/gridss-2.0.0-gridss-jar-with-dependencies.jar";
     sha256 = "01srl3qvv060whqg1y1fpxjc5cwga5wscs1bmf1v3z87dignra7k";
   };
+  gridssConfig = callBionix ./gridss-configFile.nix {};
   callVariants = callBionix ./gridss-callVariants.nix;
   computeSamTags = callBionix ./gridss-computeSamTags.nix;
   softClipsToSplitReads = callBionix ./gridss-softClipsToSplitReads.nix;
diff --git a/tools/kallisto-index.nix b/tools/kallisto-index.nix
new file mode 100644
index 0000000..33dfb80
--- /dev/null
+++ b/tools/kallisto-index.nix
@@ -0,0 +1,22 @@
+{bionix
+, nixpkgs
+, kmerSize ? 31
+, unique ? false}:
+
+with nixpkgs;
+with lib;
+with bionix.types;
+
+assert (kmerSize > 1);
+
+input:
+
+assert (matchFiletype input { fa = _: true; } input);
+
+stdenv.mkDerivation {
+  name = "kallisto-index";
+  buildInputs = [ kallisto ];
+  buildCommand = ''
+    kallisto index -k ${toString kmerSize} ${optionalString unique "--make-unique"} -i $out ${input}
+  '';
+}
diff --git a/tools/kallisto-quant.nix b/tools/kallisto-quant.nix
new file mode 100644
index 0000000..c410721
--- /dev/null
+++ b/tools/kallisto-quant.nix
@@ -0,0 +1,48 @@
+{bionix
+, nixpkgs
+, indexFlags ? {}
+, bias ? false
+, bootstrapSamples ? 0
+, seed ? 42
+, plaintext ? false
+, fusion ? false
+, single ? false
+, frStranded ? false
+, rfStranded ? false
+, fragmentLength ? null
+, fragmentSD ? null
+, ref}:
+
+with nixpkgs;
+with lib;
+
+assert (!single || (fragmentLength != null && fragmentSD != null));
+
+inputs:
+
+let
+  inherit (bionix.types) matchFiletype';
+  isFastQ = matchFiletype' "kallisto-quant" {fq = _: true; gz = isFastQ; };
+in
+
+assert (all (x: isFastQ (x.filetype)) inputs);
+
+stdenv.mkDerivation {
+  name = "kallisto-quant";
+  buildInputs = [ kallisto ];
+  buildCommand = ''
+    mkdir $out
+    kallisto quant \
+      -i ${bionix.kallisto.index indexFlags ref} \
+      -o $out \
+      ${optionalString bias "--bias"} \
+      ${optionalString (bootstrapSamples > 0) "-b ${toString bootstrapSamples} --seed=${toString seed}"} \
+      ${optionalString plaintext "--plaintext"} \
+      ${optionalString fusion "--fusion"} \
+      ${optionalString single "--single -l ${toString fragmentLength} -s ${toString fragmentSD}"} \
+      ${optionalString frStranded "--fr-stranded"} \
+      ${optionalString rfStranded "--rf-stranded"} \
+      -t $NIX_BUILD_CORES \
+      ${concatStringsSep " " inputs}
+  '';
+}
diff --git a/tools/kallisto.nix b/tools/kallisto.nix
new file mode 100644
index 0000000..f13f493
--- /dev/null
+++ b/tools/kallisto.nix
@@ -0,0 +1,8 @@
+{bionix, nixpkgs}:
+
+with bionix;
+
+{
+  index = callBionix ./kallisto-index.nix;
+  quant = callBionix ./kallisto-quant.nix;
+}
diff --git a/tools/samtools-sort.nix b/tools/samtools-sort.nix
index ab9d603..5a02dab 100644
--- a/tools/samtools-sort.nix
+++ b/tools/samtools-sort.nix
@@ -11,7 +11,7 @@ with nixpkgs;
 with lib;
 
 let
-  inherit (bionix.types) matchFiletype coordSort;
+  inherit (bionix.types) matchFiletype coordSort matchFileSorting;
 in
 
 assert (matchFiletype "samtools-sort" { bam = _: true; sam = _: true; cram = _: true; } input);
@@ -19,11 +19,20 @@ 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; } input;
 in stdenv.mkDerivation {
   name = "samtools-sort";
   buildInputs = [ samtools ];
-  buildCommand = ''
-    samtools sort -@ $NIX_BUILD_CORES ${optionalString nameSort "-n"} ${outFmtFlags} ${optionalString (flags != null) flags} ${input} > $out
-  '';
+  buildCommand =
+    if alreadySorted then
+      "ln -s $out ${input}"
+    else
+      ''
+        samtools sort -@ $NIX_BUILD_CORES \
+          ${optionalString nameSort "-n"} \
+          ${outFmtFlags} \
+          ${optionalString (flags != null) flags} \
+          ${input} > $out
+      '';
   passthru.filetype = if nameSort then bionix.types.nameSort outfmtR else coordSort outfmtR;
 }