gridss: refactor identify & annotate variants

4 files changed, 114 insertions, 169 deletions

diff --git a/tools/gridss-annotateVariants.nix b/tools/gridss-annotateVariants.nix
deleted file mode 100644
index 3560f77..0000000
--- a/tools/gridss-annotateVariants.nix
+++ /dev/null
@@ -1,85 +0,0 @@
-{ bionix
-, nixpkgs
-, bwaIndexAttrs ? {}
-, faidxAttrs ? {}
-, indexAttrs ? {}
-, assemblyAttrs ? {}
-, collectMetricsAttrs ? {}
-, softClipsToSplitReadsAttrs ? { flags = "REALIGN_ENTIRE_READ=true"; }
-, identifyVariantsAttrs ? {}
-, flags ? null
-, config ? null
-}:
-
-with nixpkgs;
-with lib;
-with bionix.types;
-with bionix.gridss;
-
-inputs:
-
-let
-  getref = matchFiletype "gridss-annotateVariants" { bam = x: x.ref; };
-  ref = getref (head inputs);
-  sorted = matchFileSorting "gridss-annotateVariants" { coord = _: true; };
-  homoRef = length (unique (map getref inputs)) == 1;
-
-  linkInput = f: attrs: input: ''
-    BASENAME=$(basename ${input})
-    WRKDIR="''${BASENAME}.gridss.working"
-    if [[ ! -e $WRKDIR ]] ; then
-      mkdir $WRKDIR
-    fi
-    for f in ${f attrs input}/* ; do
-      ln -s $f $WRKDIR/$BASENAME.''${f##*.}
-    done
-  '';
-
-  linkSV = input: ''
-    BASENAME=$(basename ${input})
-    WRKDIR="''${BASENAME}.gridss.working"
-    if [[ ! -e $WRKDIR ]] ; then
-    mkdir $WRKDIR
-    fi
-    ln -s ${input} $WRKDIR/$BASENAME.sv.bam
-    ln -s ${bionix.samtools.index indexAttrs input} $WRKDIR/$BASENAME.sv.bai
-  '';
-
-  assembly = bionix.samtools.sort {} (softClipsToSplitReads softClipsToSplitReadsAttrs (bionix.gridss.assemble assemblyAttrs inputs));
-in
-
-assert (all sorted inputs);
-assert (homoRef);
-
-stdenv.mkDerivation rec {
-  name = "gridss-identifyVariants";
-  buildInputs = [ jre ];
-  buildCommand = ''
-    ln -s ${ref} ref.fa
-    ln -s ${bionix.samtools.faidx faidxAttrs ref} ref.fa.fai
-    for f in ${bionix.bwa.index bwaIndexAttrs ref}/*; do
-      ln -s $f
-    done
-    ${concatMapStringsSep "\n" (linkSV) inputs}
-    ${linkSV assembly}
-    ${concatMapStringsSep "\n" (linkInput collectMetrics collectMetricsAttrs) inputs}
-    ${linkInput collectMetrics collectMetricsAttrs assembly}
-    ln -s ${identifyVariants identifyVariantsAttrs inputs} input.vcf
-	  java -Xmx4g -Dsamjdk.create_index=true \
-      -cp ${jar} gridss.AnnotateVariants \
-      REFERENCE_SEQUENCE=ref.fa \
-      ${concatMapStringsSep " " (i: "INPUT='${i}'") inputs} \
-      ASSEMBLY=${assembly} \
-      INPUT_VCF=input.vcf \
-      OUTPUT_VCF=out.vcf \
-      WORKING_DIR=$TMPDIR/ \
-      ${optionalString (config != null) ("OPTIONS_FILE=" +  bionix.gridss.gridssConfig config)} \
-      TMP_DIR=$TMPDIR/
-
-    mv out.vcf $out
-    '';
-  passthru = {
-    filetype = filetype.vcf { ref = ref; };
-    gridss.assembly = assembly;
-  };
-}
diff --git a/tools/gridss-identifyVariants.nix b/tools/gridss-identifyVariants.nix
deleted file mode 100644
index 23a9d85..0000000
--- a/tools/gridss-identifyVariants.nix
+++ /dev/null
@@ -1,82 +0,0 @@
-{ bionix
-, nixpkgs
-, bwaIndexAttrs ? {}
-, faidxAttrs ? {}
-, indexAttrs ? {}
-, assemblyAttrs ? {}
-, collectMetricsAttrs ? {}
-, softClipsToSplitReadsAttrs ? {}
-, flags ? null
-, config ? null
-}:
-
-with nixpkgs;
-with lib;
-with bionix.types;
-with bionix.gridss;
-
-inputs:
-
-let
-  getref = matchFiletype "gridss-identifyVariants" { bam = x: x.ref; };
-  ref = getref (head inputs);
-  sorted = matchFileSorting "gridss-identifyVariants" { coord = _: true; };
-  homoRef = length (unique (map getref inputs)) == 1;
-
-  linkInput = f: attrs: input: ''
-    BASENAME=$(basename ${input})
-    WRKDIR="''${BASENAME}.gridss.working"
-    if [[ ! -e $WRKDIR ]] ; then
-      mkdir $WRKDIR
-    fi
-    for f in ${f attrs input}/* ; do
-      ln -s $f $WRKDIR/$BASENAME.''${f##*.}
-    done
-  '';
-
-  linkSV = input: ''
-    BASENAME=$(basename ${input})
-    WRKDIR="''${BASENAME}.gridss.working"
-    if [[ ! -e $WRKDIR ]] ; then
-      mkdir $WRKDIR
-    fi
-    ln -s ${input} $WRKDIR/$BASENAME.sv.bam
-    ln -s ${bionix.samtools.index indexAttrs input} $WRKDIR/$BASENAME.sv.bai
-  '';
-
-  assembly = bionix.samtools.sort {} (softClipsToSplitReads softClipsToSplitReadsAttrs (bionix.samtools.sort { nameSort = true;} (bionix.gridss.assemble assemblyAttrs inputs)));
-in
-
-assert (all sorted inputs);
-assert (homoRef);
-
-stdenv.mkDerivation rec {
-  name = "gridss-identifyVariants";
-  buildInputs = [ jre samtools ];
-  buildCommand = ''
-    ln -s ${ref} ref.fa
-    ln -s ${bionix.samtools.faidx faidxAttrs ref} ref.fa.fai
-    for f in ${bionix.bwa.index bwaIndexAttrs ref}/*; do
-      ln -s $f
-    done
-    ${concatMapStringsSep "\n" (linkSV) inputs}
-    ${linkSV assembly}
-    ${concatMapStringsSep "\n" (linkInput collectMetrics collectMetricsAttrs) inputs}
-    ${linkInput collectMetrics collectMetricsAttrs assembly}
-	  java -Xmx4g -Dsamjdk.create_index=true \
-      -cp ${jar} gridss.IdentifyVariants \
-      REFERENCE_SEQUENCE=ref.fa \
-      ${concatMapStringsSep " " (i: "INPUT='${i}'") inputs} \
-      ASSEMBLY=${assembly} \
-      OUTPUT_VCF=out.vcf \
-      ${optionalString (config != null) ("OPTIONS_FILE=" + bionix.gridss.gridssConfig config)} \
-      WORKING_DIR=$TMPDIR/ \
-      TMP_DIR=$TMPDIR/
-
-    mv out.vcf $out
-    '';
-  passthru = {
-    filetype = filetype.vcf { ref = ref; };
-    gridss.assembly = assembly;
-  };
-}
diff --git a/tools/gridss-variants.nix b/tools/gridss-variants.nix
new file mode 100644
index 0000000..a838da6
--- /dev/null
+++ b/tools/gridss-variants.nix
@@ -0,0 +1,112 @@
+{ bionix
+, nixpkgs
+, bwaIndexAttrs ? {}
+, faidxAttrs ? {}
+, indexAttrs ? {}
+, assemblyAttrs ? {}
+, collectMetricsAttrs ? {}
+, softClipsToSplitReadsAttrs ? { flags = "REALIGN_ENTIRE_READ=true"; }
+, flags ? null
+, config ? null
+}:
+
+with nixpkgs;
+with lib;
+with bionix.types;
+with bionix.gridss;
+
+inputs:
+
+let
+  getref = matchFiletype "gridss-variants" { bam = x: x.ref; };
+  ref = getref (head inputs);
+  sorted = matchFileSorting "gridss-variants" { coord = _: true; };
+  homoRef = length (unique (map getref inputs)) == 1;
+
+  linkInput = f: attrs: input: ''
+    BASENAME=$(basename ${input})
+    WRKDIR="''${BASENAME}.gridss.working"
+    if [[ ! -e $WRKDIR ]] ; then
+      mkdir $WRKDIR
+    fi
+    for f in ${f attrs input}/* ; do
+      ln -s $f $WRKDIR/$BASENAME.''${f##*.}
+    done
+  '';
+
+  linkSV = input: ''
+    BASENAME=$(basename ${input})
+    WRKDIR="''${BASENAME}.gridss.working"
+    if [[ ! -e $WRKDIR ]] ; then
+      mkdir $WRKDIR
+    fi
+    ln -s ${input} $WRKDIR/$BASENAME.sv.bam
+    ln -s ${bionix.samtools.index indexAttrs input} $WRKDIR/$BASENAME.sv.bai
+  '';
+
+  assembly = bionix.samtools.sort {} (softClipsToSplitReads softClipsToSplitReadsAttrs (bionix.gridss.assemble assemblyAttrs inputs));
+
+  mkLinks = ''
+    ln -s ${ref} ref.fa
+    ln -s ${bionix.samtools.faidx faidxAttrs ref} ref.fa.fai
+    for f in ${bionix.bwa.index bwaIndexAttrs ref}/*; do
+      ln -s $f
+    done
+    ${concatMapStringsSep "\n" (linkSV) inputs}
+    ${linkSV assembly}
+    ${concatMapStringsSep "\n" (linkInput collectMetrics collectMetricsAttrs) inputs}
+    ${linkInput collectMetrics collectMetricsAttrs assembly}
+  '';
+
+in
+
+assert (all sorted inputs);
+assert (homoRef);
+
+rec {
+  identify = stdenv.mkDerivation rec {
+    name = "gridss-identifyVariants";
+    buildInputs = [ jre samtools ];
+    buildCommand = mkLinks + ''
+      java -Xmx4g -Dsamjdk.create_index=true \
+        -cp ${jar} gridss.IdentifyVariants \
+        REFERENCE_SEQUENCE=ref.fa \
+        ${concatMapStringsSep " " (i: "INPUT='${i}'") inputs} \
+        ASSEMBLY=${assembly} \
+        OUTPUT_VCF=out.vcf \
+        ${optionalString (config != null) ("OPTIONS_FILE=" + bionix.gridss.gridssConfig config)} \
+        WORKING_DIR=$TMPDIR/ \
+        TMP_DIR=$TMPDIR/
+
+      mv out.vcf $out
+      '';
+    passthru = {
+      filetype = filetype.vcf { ref = ref; };
+      gridss.assembly = assembly;
+    };
+  };
+
+  annotate = stdenv.mkDerivation rec {
+    name = "gridss-annotateVariants";
+    buildInputs = [ jre ];
+    buildCommand = mkLinks + ''
+      ln -s ${identify} input.vcf
+      java -Xmx4g -Dsamjdk.create_index=true \
+        -cp ${jar} gridss.AnnotateVariants \
+        REFERENCE_SEQUENCE=ref.fa \
+        ${concatMapStringsSep " " (i: "INPUT='${i}'") inputs} \
+        ASSEMBLY=${assembly} \
+        INPUT_VCF=input.vcf \
+        OUTPUT_VCF=out.vcf \
+        WORKING_DIR=$TMPDIR/ \
+        ${optionalString (config != null) ("OPTIONS_FILE=" +  bionix.gridss.gridssConfig config)} \
+        TMP_DIR=$TMPDIR/
+
+      mv out.vcf $out
+      '';
+    passthru = {
+      filetype = filetype.vcf { ref = ref; };
+      gridss.assembly = assembly;
+    };
+  };
+}
diff --git a/tools/gridss.nix b/tools/gridss.nix
index b85e84e..d47a669 100644
--- a/tools/gridss.nix
+++ b/tools/gridss.nix
@@ -15,8 +15,8 @@ rec {
   collectMetrics = callBionix ./gridss-collectMetrics.nix;
   extractSVReads = callBionix ./gridss-extractSVReads.nix;
   assemble = callBionix ./gridss-assemble.nix;
-  identifyVariants = callBionix ./gridss-identifyVariants.nix;
-  annotateVariants = callBionix ./gridss-annotateVariants.nix;
+  identifyVariants = attrs: input: ((callBionix ./gridss-variants.nix attrs) input).identify;
+  annotateVariants = attrs: input: ((callBionix ./gridss-variants.nix attrs) input).annotate;
   preprocessBam = input: with samtools; sort {} (gridss.softClipsToSplitReads {} (gridss.computeSamTags {} (sort {nameSort = true;} (gridss.extractSVReads {} (markdup {} (sort {} (fixmate {mateScore = true;} (sort {nameSort = true;} input))))))));
   call = inputs: bionix.gridss.annotateVariants {} (map gridss.preprocessBam inputs);
 }