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

8 files changed, 50 insertions, 26 deletions

diff --git a/default.nix b/default.nix
index f835fa0..50b85a3 100644
--- a/default.nix
+++ b/default.nix
@@ -1,6 +1,8 @@
 {nixpkgs ? import <nixpkgs> {}}:
 
 let
+  inherit (nixpkgs) fetchurl;
+
   bionix = nixpkgs.lib.makeExtensible (self:
   let callBionix = file: attrs: import file ({ bionix = self; nixpkgs = nixpkgs; } // attrs);
   in with self; {
@@ -29,5 +31,12 @@ let
     def = f: defs: attrs: f (defs // attrs);
     defQsub = qsubAttrs: f: defs: qsubAttr qsubAttrs (def f defs);
 
+    # Fetching files of specific type
+    fetchFastQ = attrs: with types; tagFiletype (filetype.fq {}) (fetchurl attrs);
+    fetchFastA = attrs: with types; tagFiletype (filetype.fa {}) (fetchurl attrs);
+    fetchFastQGZ = attrs: with types; tagFiletype (filetype.gz (filetype.fq {})) (fetchurl attrs);
+    fetchFastAGZ = attrs: with types; tagFiletype (filetype.gz (filetype.fa {})) (fetchurl attrs);
+
+
   });
 in bionix
diff --git a/tools/gridss-annotateVariants.nix b/tools/gridss-annotateVariants.nix
index 6c8fbc6..122c451 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 ${jar} gridss.AnnotateVariants \
@@ -65,7 +73,7 @@ stdenv.mkDerivation rec {
       INPUT_VCF=input.vcf \
       OUTPUT_VCF=out.vcf \
       WORKING_DIR=$TMPDIR/ \
-      ${optionalString (config != null) ("CONFIGURATION_FILE=" +  bionix.gridss.ggridssConfig config)} \
+      ${optionalString (config != null) ("CONFIGURATION_FILE=" +  bionix.gridss.gridssConfig config)} \
       TMP_DIR=$TMPDIR/
 
     mv out.vcf $out
diff --git a/tools/gridss-assemble.nix b/tools/gridss-assemble.nix
index 1e25731..d1a3e4b 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 54edc85..b50a01c 100644
--- a/tools/gridss-extractSVReads.nix
+++ b/tools/gridss-extractSVReads.nix
@@ -5,6 +5,7 @@
 , flags ? null
 , unmappedReads ? false
 , minClipLength ? 5
+, collectMetricsAttrs ? {}
 , config ? null
 }:
 
@@ -27,16 +28,17 @@ stdenv.mkDerivation rec {
     ln -s ${bionix.samtools.faidx faidxAttrs ref} ref.fa.fai
     ln -s ${bionix.samtools.dict dictIndexAttrs ref} ref.fa.dict
     ln -s ${input} input.bam
-    mkdir $out
+    for f in ${bionix.gridss.collectMetrics collectMetricsAttrs input}/* ; do
+      ln -s $f
+    done
     java -Dsamjdk.create_index=true \
       -cp ${bionix.gridss.jar} gridss.ExtractSVReads \
       REFERENCE_SEQUENCE=ref.fa \
       I=input.bam \
-      O=$out/input.sv.bam \
-      METRICS_OUTPUT=$out/input.sv_metrics \
-      INSERT_SIZE_METRICS=$out/input.insert_size_metrics \
+      O=$out \
       UNMAPPED_READS=${if unmappedReads then "true" else "false"} \
       ${optionalString (config != null) ("CONFIGURATION_FILE=" + bionix.gridss.gridssConfig config)} \
       MIN_CLIP_LENGTH=${toString minClipLength}
   '';
+  passthru.filetype = input.filetype;
 }
diff --git a/tools/gridss-identifyVariants.nix b/tools/gridss-identifyVariants.nix
index a53fcb7..e5c74a4 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
 
@@ -42,19 +52,17 @@ assert (homoRef);
 
 stdenv.mkDerivation rec {
   name = "gridss-identifyVariants";
-  buildInputs = [ jre ];
+  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" (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
 	  java -Xmx4g -Dsamjdk.create_index=true \
       -cp ${jar} gridss.IdentifyVariants \
       REFERENCE_SEQUENCE=ref.fa \
diff --git a/tools/gridss-softClipsToSplitReads.nix b/tools/gridss-softClipsToSplitReads.nix
index 47a3abd..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 ];
@@ -37,5 +35,5 @@ stdenv.mkDerivation rec {
       ${optionalString (config != null) ("CONFIGURATION_FILE=" + bionix.gridss.gridssConfig config)} \
 			WORKER_THREADS=$NIX_BUILD_CORES
     '';
-  passthru.filetype = filetype.bam { ref = ref; sorting = sort.none {}; };
+  passthru.filetype = filetype.bam { ref = ref; sorting = matchFileSorting "grids-softClipsToSplitReads" { coord = _: input.sorting; name = _: sort.none {}; none = _: input.sorting;} input;};
 }
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);
 }
diff --git a/tools/samtools-sort.nix b/tools/samtools-sort.nix
index 5a02dab..e77f3db 100644
--- a/tools/samtools-sort.nix
+++ b/tools/samtools-sort.nix
@@ -19,7 +19,7 @@ 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;
+  alreadySorted = matchFileSorting "samtools-sort" { name = _: nameSort; coord = _: !nameSort; none = _: false;} input;
 in stdenv.mkDerivation {
   name = "samtools-sort";
   buildInputs = [ samtools ];