gridss: restructure gridss pipeline

6 files changed, 30 insertions, 14 deletions

diff --git a/tools/gridss-annotateVariants.nix b/tools/gridss-annotateVariants.nix
index 4422cd6..cae5daf 100644
--- a/tools/gridss-annotateVariants.nix
+++ b/tools/gridss-annotateVariants.nix
@@ -2,8 +2,8 @@
 , nixpkgs
 , bwaIndexAttrs ? {}
 , faidxAttrs ? {}
+, indexAttrs ? {}
 , assemblyAttrs ? {}
-, extractSVReadsAttrs ? {}
 , collectMetricsAttrs ? {}
 , softClipsToSplitReadsAttrs ? {}
 , identifyVariantsAttrs ? {}
@@ -35,6 +35,16 @@ let
     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
 
@@ -50,12 +60,10 @@ stdenv.mkDerivation rec {
     for f in ${bionix.bwa.index bwaIndexAttrs ref}/*; do
       ln -s $f
     done
-    ${concatMapStringsSep "\n" (linkInput extractSVReads extractSVReadsAttrs) inputs}
+    ${concatMapStringsSep "\n" (linkSV) inputs}
+    ${linkSV assembly}
     ${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 \
diff --git a/tools/gridss-assemble.nix b/tools/gridss-assemble.nix
index cdda748..1bda91f 100644
--- a/tools/gridss-assemble.nix
+++ b/tools/gridss-assemble.nix
@@ -2,8 +2,8 @@
 , nixpkgs
 , bwaIndexAttrs ? {}
 , faidxAttrs ? {}
+, indexAttrs ? {}
 , collectMetricsAttrs ? {}
-, extractSVReadsAttrs ? {}
 , flags ? null
 }:
 
@@ -23,9 +23,8 @@ let
     BASENAME=$(basename ${input})
     WRKDIR="''${BASENAME}.gridss.working"
     mkdir $WRKDIR
-    for f in ${bionix.gridss.extractSVReads extractSVReadsAttrs input}/* ; do
-      ln -s $f $WRKDIR/$BASENAME.''${f#*.}
-    done
+    ln -s ${input} $WRKDIR/$BASENAME.sv.bam
+    ln -s  ${bionix.samtools.index indexAttrs input} $WRKDIR/$BASENAME.sv.bai
     for f in ${bionix.gridss.collectMetrics collectMetricsAttrs input}/* ; do
       ln -s $f $WRKDIR/$BASENAME.''${f#*.}
     done
diff --git a/tools/gridss-extractSVReads.nix b/tools/gridss-extractSVReads.nix
index b5ceffb..d260086 100644
--- a/tools/gridss-extractSVReads.nix
+++ b/tools/gridss-extractSVReads.nix
@@ -5,6 +5,7 @@
 , flags ? null
 , unmappedReads ? false
 , minClipLength ? 5
+, collectMetricsAttrs ? {}
 , config ? null
 }:
 
diff --git a/tools/gridss-identifyVariants.nix b/tools/gridss-identifyVariants.nix
index c91de00..d37e7ae 100644
--- a/tools/gridss-identifyVariants.nix
+++ b/tools/gridss-identifyVariants.nix
@@ -2,8 +2,8 @@
 , nixpkgs
 , bwaIndexAttrs ? {}
 , faidxAttrs ? {}
+, indexAttrs ? {}
 , assemblyAttrs ? {}
-, extractSVReadsAttrs ? {}
 , collectMetricsAttrs ? {}
 , softClipsToSplitReadsAttrs ? {}
 , flags ? null
@@ -34,6 +34,16 @@ let
     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
 
diff --git a/tools/gridss-softClipsToSplitReads.nix b/tools/gridss-softClipsToSplitReads.nix
index 1f9c8b4..2fff15c 100644
--- a/tools/gridss-softClipsToSplitReads.nix
+++ b/tools/gridss-softClipsToSplitReads.nix
@@ -17,8 +17,6 @@ let
   ref = matchFiletype "gridss-softClipsToSplitReads" { bam = x: x.ref; } input;
 in
 
-assert (matchFileSorting "gridss-softClipsToSplitReads" { name = _: true; } input);
-
 stdenv.mkDerivation rec {
   name = "gridss-softClipsToSplitReads";
   buildInputs = [ jre bwa ];
diff --git a/tools/gridss.nix b/tools/gridss.nix
index 1e4e2d1..b85e84e 100644
--- a/tools/gridss.nix
+++ b/tools/gridss.nix
@@ -17,6 +17,6 @@ rec {
   assemble = callBionix ./gridss-assemble.nix;
   identifyVariants = callBionix ./gridss-identifyVariants.nix;
   annotateVariants = callBionix ./gridss-annotateVariants.nix;
-  preprocessBam = input: with samtools; markdup {} (sort {} (fixmate {mateScore = true;} (bionix.gridss.softClipsToSplitReads {} (bionix.gridss.computeSamTags {} (sort {nameSort = true;} input)))));
-  call = inputs: bionix.gridss.annotateVariants {} (map bionix.gridss.preprocessBam inputs);
+  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);
 }