From e7cd661d1c5fb4135e3d436e151294e26aef9127 Mon Sep 17 00:00:00 2001
From: Justin Bedo <cu@cua0.org>
Date: Mon, 29 Oct 2018 15:33:53 +1100
Subject: Split gridss into constituents

Wrap each individual command for GRIDSS so that bionix executed the pipeline
rather than GRIDSS. This patch introduces a "call" function that executed the
whole pipeline in bionix on an arbitrary BAM file.

Resolves #10.
---
 tools/gridss-annotateVariants.nix | 75 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 75 insertions(+)
 create mode 100644 tools/gridss-annotateVariants.nix

(limited to 'tools/gridss-annotateVariants.nix')

diff --git a/tools/gridss-annotateVariants.nix b/tools/gridss-annotateVariants.nix
new file mode 100644
index 0000000..4f66c6c
--- /dev/null
+++ b/tools/gridss-annotateVariants.nix
@@ -0,0 +1,75 @@
+{ bionix
+, nixpkgs
+, bwaIndexAttrs ? {}
+, faidxAttrs ? {}
+, assemblyAttrs ? {}
+, extractSVReadsAttrs ? {}
+, collectMetricsAttrs ? {}
+, softClipsToSplitReadsAttrs ? {}
+, identifyVariantsAttrs ? {}
+, flags ? 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
+  '';
+
+  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 ];
+  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" (linkInput extractSVReads extractSVReadsAttrs) inputs}
+    ${concatMapStringsSep "\n" (linkInput collectMetrics collectMetricsAttrs) inputs}
+    ${linkInput collectMetrics collectMetricsAttrs assembly}
+    ASSBASE=$(basename ${assembly})
+    ln -s ${assembly} $ASSBASE.gridss.working/$ASSBASE.sv.bam
+    ln -s ${bionix.samtools.index {} assembly} $ASSBASE.gridss.working/$ASSBASE.sv.bai
+    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/ \
+      TMP_DIR=$TMPDIR/
+
+    mv out.vcf $out
+    '';
+  passthru = {
+    filetype = filetype.vcf { ref = ref; };
+    gridss.assembly = assembly;
+  };
+}
-- 
cgit v1.2.3