From 36a0a03eda2ec8cf9b1f5fea8c437b68638d1de5 Mon Sep 17 00:00:00 2001
From: Justin Bedo <cu@cua0.org>
Date: Mon, 26 Sep 2022 11:51:52 +1000
Subject: samtools.view,mosdepth.depth: add targets parameter

---
 test-tnpair.nix          | 159 +++++++++++++++++++++++++++++------------------
 tools/mosdepth-depth.nix |  51 +++++++++------
 tools/samtools-view.nix  |  80 +++++++++++++++++-------
 3 files changed, 190 insertions(+), 100 deletions(-)

diff --git a/test-tnpair.nix b/test-tnpair.nix
index 195ac27..ac8d0bb 100644
--- a/test-tnpair.nix
+++ b/test-tnpair.nix
@@ -1,25 +1,32 @@
-with import <bionix> { };
-with lib;
-
-let
-  fetchlocal = path: pkgs.stdenv.mkDerivation {
-    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);
+with import <bionix> {};
+with lib; let
+  fetchlocal = path:
+    pkgs.stdenv.mkDerivation {
+      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);
 
   ref = fetchfa ./examples/ref.fa;
 
-  alignWithRG = aligner: rg: x: nameSort (aligner { inherit ref; RG = { ID = rg; SM = 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 { };
+  alignWithRG = aligner: rg: x:
+    nameSort (aligner {
+        inherit ref;
+        RG = {
+          ID = rg;
+          SM = 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 = {
@@ -38,42 +45,56 @@ let
     };
   };
 
-  processPair = { tumour, normal }: rec {
-    alignments = mapAttrs (_: x: markdup (sort (fixmate (alignWithRG bwa.align x.name x.files)))) { inherit normal tumour; };
-    bowtie-alignments = mapAttrs (_: x: markdup (sort (fixmate (alignWithRG bowtie.align x.name x.files)))) { inherit normal tumour; };
-    bwa2-alignments = mapAttrs (_: x: markdup (sort (fixmate (alignWithRG bwa.mem2 x.name x.files)))) { inherit normal tumour; };
+  processPair = {
+    tumour,
+    normal,
+  }: rec {
+    alignments = mapAttrs (_: x: markdup (sort (fixmate (alignWithRG bwa.align x.name x.files)))) {inherit normal tumour;};
+    bowtie-alignments = mapAttrs (_: x: markdup (sort (fixmate (alignWithRG bowtie.align x.name x.files)))) {inherit normal tumour;};
+    bwa2-alignments = mapAttrs (_: x: markdup (sort (fixmate (alignWithRG bwa.mem2 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;
-    "subread.bam" = subread.align { inherit ref; } tnpair.normal.files;
-    "last.maf" = lastal.align { inherit ref; } tnpair.normal.files.input1;
-    "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;
+    "subread.bam" = subread.align {inherit ref;} tnpair.normal.files;
+    "last.maf" = lastal.align {inherit ref;} tnpair.normal.files.input1;
+    "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; } (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;} (with tnpairResult.alignments; [normal tumour]);
     cnvkit = cnvkitResults.cnvs;
     "cnvkit.pdf" = cnvkitResults.plot;
     "octopus.vcf" = tnpairResult.octopusVars;
@@ -81,27 +102,45 @@ 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 ]);
-    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 ]);
-    gridssSomatic = gridss.filterSomatic { normalName = "mysample2"; } (gridss.call (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) ];
-    "${tnpair.tumour.name}.cram" = samtools.view { outfmt = types.toCram; } tnpairResult.alignments.tumour;
+    delly = delly.call {} (with tnpairResult.alignments; [normal tumour]);
+    manta = manta.call {} (with tnpairResult.alignments; {normals = [normal tumour];});
+    mantaTN = manta.call {} (with tnpairResult.alignments; {
+      normals = [normal];
+      inherit tumour;
+    });
+    mantaT = manta.call {} (with tnpairResult.alignments; {inherit tumour;});
+    gridss = gridss.callVariants {} (with tnpairResult.alignments; [normal tumour]);
+    gridss2 = gridss.call (with tnpairResult.alignments; [normal tumour]);
+    gridss3 = gridss.callAndAssemble (with tnpairResult.alignments; [normal tumour]);
+    gridssSomatic = gridss.filterSomatic {normalName = "mysample2";} (gridss.call (with tnpairResult.alignments; [normal tumour]));
+    "merged-shards.bam" = samtools.merge {} tnpairResult.shards;
+    "merged.bam" = samtools.merge {} [tnpairResult.alignments.tumour tnpairResult.alignments.normal];
+    "merged-namesorted.bam" = samtools.merge {} [(nameSort tnpairResult.alignments.tumour) (nameSort tnpairResult.alignments.normal)];
+    "${tnpair.tumour.name}.cram" = samtools.view {outfmt = types.toCram;} tnpairResult.alignments.tumour;
+    "${tnpair.tumour.name}-targeted.cram" =
+      samtools.view {
+        outfmt = types.toCram;
+        targets = "abcd\t10\t100";
+      }
+      tnpairResult.alignments.tumour;
     "${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"];
+    };
+    mosdepth-targeted = mosdepth.plot {} {
+      inputs = mapAttrsToList (_: mosdepth.depth {targets = ["abc\t10\t100"];}) tnpairResult.alignments;
+      names = ["seq1" "seq2"];
+    };
+    xenomapper = xenomapper.allocate {} {
+      primary = nameSort tnpairResult.alignments.tumour;
+      secondary = nameSort tnpairResult.alignments.tumour;
+    };
     quip = with quip; unquip {} (quip {} tnpairResult.alignments.tumour);
-    genmap = genmap.calcmap { } ref;
+    genmap = genmap.calcmap {} ref;
     inherit alignments;
   };
-
 in
-testNaming
+  testNaming
diff --git a/tools/mosdepth-depth.nix b/tools/mosdepth-depth.nix
index 2a72989..4627071 100644
--- a/tools/mosdepth-depth.nix
+++ b/tools/mosdepth-depth.nix
@@ -1,21 +1,36 @@
-{ bionix
-, indexAttrs ? { }
-, flags ? null
+{
+  bionix,
+  indexAttrs ? {},
+  targets ? null,
+  flags ? null,
 }:
-
 with bionix;
 with lib;
-
-input:
-
-stage {
-  name = "mosdepth-depth";
-  buildInputs = with pkgs; [ mosdepth ];
-  buildCommand = ''
-    mkdir $out
-    ln -s ${input} input.bam
-    ln -s ${bionix.samtools.index indexAttrs input} input.bam.bai
-    mosdepth -t $NIX_BUILD_CORES ${optionalString (flags != null) flags} $out/out input.bam
-  '';
-  passthru.multicore = true;
-}
+  input: let
+    handleTarget = x: let
+      type = builtins.typeOf x;
+      handler = handlers."${type}" or (builtins.throw "mosdepth-depth:unhandled target type:${type}");
+      handlers = {
+        string = handleTarget [x];
+        list = let
+          file = pkgs.writeText "target.bed" (concatStringsSep "\n" x);
+        in "-b ${file}";
+        path = "-b ${x}";
+        set = "-b ${x}";
+      };
+    in
+      handler;
+  in
+    stage {
+      name = "mosdepth-depth";
+      buildInputs = with pkgs; [mosdepth];
+      buildCommand = ''
+        mkdir $out
+        ln -s ${input} input.bam
+        ln -s ${bionix.samtools.index indexAttrs input} input.bam.bai
+        mosdepth -t $NIX_BUILD_CORES \
+          ${optionalString (targets != null) (handleTarget targets)} \
+          ${optionalString (flags != null) flags} $out/out input.bam
+      '';
+      passthru.multicore = true;
+    }
diff --git a/tools/samtools-view.nix b/tools/samtools-view.nix
index 4d2bdc7..4966b13 100644
--- a/tools/samtools-view.nix
+++ b/tools/samtools-view.nix
@@ -1,27 +1,63 @@
-{ bionix
-, nameSort ? false
-, flags ? null
-, outfmt ? null
-}:
-
-input:
-
+{
+  bionix,
+  nameSort ? false,
+  flags ? null,
+  outfmt ? null,
+  targets ? null,
+}: input:
 with bionix;
 with lib;
 with types;
+assert (matchFiletype "samtools-view" {
+    bam = _: true;
+    sam = _: true;
+    cram = _: true;
+  }
+  input); let
+  handleTarget = x: let
+    type = builtins.typeOf x;
+    handler = handlers."${type}" or (builtins.throw "samtools-view:unhandled target type:${type}");
+    handlers = {
+      string = handleTarget [x];
+      list = let
+        file = pkgs.writeText "target.bed" (concatStringsSep "\n" x);
+      in "--target-file ${file}";
+      path = "--target-file ${x}";
+      set = "--target-file ${x}";
+    };
+  in
+    handler;
 
-assert (matchFiletype "samtools-view" { bam = _: true; sam = _: true; cram = _: true; } input);
-
-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 ${fa x.ref}"; } { filetype = outfmtR; };
+  outfmtR =
+    if outfmt != null
+    then
+      (
+        if builtins.typeOf outfmt == "string"
+        then
+          {
+            "bam" = toBam;
+            "cram" = toCram;
+            "sam" = toSam;
+          }
+          ."${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 {
-  name = "samtools-view";
-  buildInputs = with pkgs; [ samtools ];
-  buildCommand = ''
-    samtools view ${outfmtFlags} ${optionalString (flags != null) flags} ${input} > $out
-  '';
-  passthru.filetype = outfmtR;
-}
+  stage {
+    name = "samtools-view";
+    buildInputs = with pkgs; [samtools];
+    buildCommand = ''
+      samtools view \
+        ${optionalString (targets != null) (handleTarget targets)} \
+        ${outfmtFlags} ${optionalString (flags != null) flags} ${input} > $out
+    '';
+    passthru.filetype = outfmtR;
+  }
-- 
cgit v1.2.3