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

10 files changed, 90 insertions, 20 deletions

diff --git a/default.nix b/default.nix
index 50b85a3..d607aa2 100644
--- a/default.nix
+++ b/default.nix
@@ -12,6 +12,7 @@ let
     types = callBionix ./lib/types.nix {};
 
     bwa = callBionix ./tools/bwa.nix {};
+    bowtie = callBionix ./tools/bowtie.nix {};
     compression = callBionix ./tools/compression.nix {};
     crumble = callBionix ./tools/crumble.nix {};
     fastqc = callBionix ./tools/fastqc.nix {};
diff --git a/lib/qsub.nix b/lib/qsub.nix
index 4ef6dbc..f7b3c3e 100644
--- a/lib/qsub.nix
+++ b/lib/qsub.nix
@@ -5,17 +5,17 @@
   args = let
     script = writeScript "qsub-script" ''
       #!${stdenv.shell}
-      while [ ! -e ${tmpDir}/$PBS_JOBID ] ; do
+      while [ ! -e ${tmpDir}/qsub-$PBS_JOBID ] ; do
         sleep ${toString sleepTime}
       done
-      TMPDIR=${tmpDir}/$PBS_JOBID
+      set -a
+      . ${tmpDir}/qsub-$PBS_JOBID/nix-set
+      set +a
+      TMPDIR=${tmpDir}/qsub-$PBS_JOBID
       TEMP=$TMPDIR
       TMP=$TMPDIR
       NIX_BUILD_TOP=$TMPDIR
       cd $TMPDIR
-      set -a
-      . nix-set
-      set +a
       ${builder} ${lib.escapeShellArgs args} > qsub-stdout 2> qsub-stderr
       echo $? > qsub-exit
     '';
@@ -46,15 +46,15 @@
       }
       trap cleanup INT TERM EXIT
 
-      cp -r $TMPDIR ${tmpDir}/$id
-      set > ${tmpDir}/$id/nix-set
+      cp -r $TMPDIR ${tmpDir}/qsub-$id
+      set > ${tmpDir}/qsub-$id/nix-set
       until qstat -f ''${id%%.} 2>&1 | grep "\(Unknown Job\|job_state = C\)" > /dev/null ; do
         sleep ${toString sleepTime}
       done
-      cat ${tmpDir}/$id/qsub-stderr >&2
-      cat ${tmpDir}/$id/qsub-stdout
-      if [ -e ${tmpDir}/$id/qsub-exit ]; then
-        exitCode=$(cat ${tmpDir}/$id/qsub-exit)
+      cat ${tmpDir}/qsub-$id/qsub-stderr >&2
+      cat ${tmpDir}/qsub-$id/qsub-stdout
+      if [ -e ${tmpDir}/qsub-$id/qsub-exit ]; then
+        exitCode=$(cat ${tmpDir}/qsub-$id/qsub-exit)
       else
         exitCode=1
       fi
diff --git a/test-tnpair.nix b/test-tnpair.nix
index a6ea163..b01d61e 100644
--- a/test-tnpair.nix
+++ b/test-tnpair.nix
@@ -21,7 +21,9 @@ let
   fetchfq = attrs: types.tagFiletype (types.filetype.fq {}) (fetchlocal attrs);
   fetchfa = attrs: types.tagFiletype (types.filetype.fa {}) (fetchlocal attrs);
 
-  alignWithRG = rg: bwa.align { ref = fetchfa ./example/ref.fa; flags = "-R'@RG\\tID:${rg}\\tSM:${rg}'";};
+  ref = fetchfa ./example/ref.fa;
+
+  alignWithRG = rg: bwa.align { inherit ref; flags = "-R'@RG\\tID:${rg}\\tSM:${rg}'";};
   sort = samtools.sort {};
   flagstat = samtools.flagstat {};
   check = fastqc.check {};
@@ -35,7 +37,7 @@ let
     };
     normal = {name = "mysample2"; files = {
         input1 = fetchfq ./example/sample2-1.fq;
-        input2 = fetchfq ./example/sample2-1.fq;
+        input2 = fetchfq ./example/sample2-2.fq;
       };
     };
   };
@@ -53,8 +55,9 @@ let
       mkdir $out
       ln -s ${tnpairResult.variants} $out/strelka
       mkdir $out/alignments
-      ln -s ${gridss.callVariants {} (with tnpairResult.alignments; [tumour])} $out/gridss
-      ln -s ${gridss.call (with tnpairResult.alignments; [tumour])} $out/gridss2
+      ln -s ${bowtie.align {inherit ref;} tnpair.normal.files} $out/alignments/bowtie-normal.bam
+      ln -s ${gridss.callVariants {} (with tnpairResult.alignments; [normal tumour])} $out/gridss
+      ln -s ${gridss.call (with tnpairResult.alignments; [normal tumour])} $out/gridss2
       ln -s ${samtools.merge {} [tnpairResult.alignments.tumour tnpairResult.alignments.normal]} $out/alignments/merged.bam
       ln -s ${samtools.view { outfmt = types.toCram; } (tnpairResult.alignments.tumour)} $out/alignments/${tnpair.tumour.name}.cram
       ln -s ${samtools.view { outfmt = types.toCram; } (tnpairResult.alignments.normal)} $out/alignments/${tnpair.normal.name}.cram
diff --git a/tools/bowtie-align.nix b/tools/bowtie-align.nix
new file mode 100644
index 0000000..456747a
--- /dev/null
+++ b/tools/bowtie-align.nix
@@ -0,0 +1,36 @@
+{ bionix
+, nixpkgs
+, ref
+, bamOutput ? true
+, flags ? null
+, indexAttrs ? {}
+}:
+
+{ input1
+, input2 ? null
+}:
+
+with nixpkgs;
+with lib;
+with bionix.types;
+with bionix.compression;
+
+let
+  fa = f: matchFiletype "bowtie2-ref" { fa = _: f; } f;
+  fq = f: matchFiletype "bowtie2-input" { fq = _: f; gz = matchFiletype' "bowtie2-input" { fq = _: f; }; } f;
+
+in stdenv.mkDerivation {
+  name = "bowtie2-align";
+  buildInputs = [ bowtie2 bc ] ++ optional bamOutput samtools;
+  buildCommand = ''
+    cores=$(echo $NIX_BUILD_CORES ${optionalString bamOutput "- 1"} | bc)
+    if [[ $cores -lt 1 ]] ; then
+      cores=1
+    fi
+    bowtie2 -x ${bionix.bowtie.index indexAttrs ref}/ref ${optionalString (flags != null) flags} --threads $cores \
+      ${if input2 != null then "-1 " + fq input1 + " -2 " + fq input2 else "-U " + fq input1} \
+      ${optionalString bamOutput "| samtools view -b"} \
+      > $out
+  '';
+  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.name {};} else filetype.sam {ref = ref; sorting = sort.name {};};
+}
diff --git a/tools/bowtie-index.nix b/tools/bowtie-index.nix
new file mode 100644
index 0000000..98e529f
--- /dev/null
+++ b/tools/bowtie-index.nix
@@ -0,0 +1,22 @@
+{ bionix
+, nixpkgs
+, flags ? null
+, seed ? 42
+}:
+
+ref:
+
+with nixpkgs;
+with lib;
+with bionix.types;
+
+assert (matchFiletype "bowtie-index" { fa = _: true; } ref);
+
+stdenv.mkDerivation {
+  name = "bowtie-index";
+  buildInputs = [ bowtie2 ];
+  buildCommand = ''
+    mkdir $out
+    bowtie2-build --seed ${toString seed} --threads $NIX_BUILD_CORES ${optionalString (flags != null) flags} ${ref} $out/ref
+  '';
+}
diff --git a/tools/bowtie.nix b/tools/bowtie.nix
new file mode 100644
index 0000000..a9fa2e8
--- /dev/null
+++ b/tools/bowtie.nix
@@ -0,0 +1,8 @@
+{ bionix, nixpkgs }:
+
+with bionix;
+
+{
+  align = callBionix ./bowtie-align.nix;
+  index = callBionix ./bowtie-index.nix;
+}
diff --git a/tools/gridss-annotateVariants.nix b/tools/gridss-annotateVariants.nix
index d5390b9..e0a6f0f 100644
--- a/tools/gridss-annotateVariants.nix
+++ b/tools/gridss-annotateVariants.nix
@@ -5,7 +5,7 @@
 , indexAttrs ? {}
 , assemblyAttrs ? {}
 , collectMetricsAttrs ? {}
-, softClipsToSplitReadsAttrs ? {}
+, softClipsToSplitReadsAttrs ? { flags = "REALIGN_ENTIRE_READ=true"; }
 , identifyVariantsAttrs ? {}
 , flags ? null
 , config ? null
@@ -47,7 +47,7 @@ let
     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)));
+  assembly = bionix.samtools.sort {} (softClipsToSplitReads softClipsToSplitReadsAttrs (bionix.gridss.assemble assemblyAttrs inputs));
 in
 
 assert (all sorted inputs);
diff --git a/tools/gridss-assemble.nix b/tools/gridss-assemble.nix
index 00408d6..f57bc12 100644
--- a/tools/gridss-assemble.nix
+++ b/tools/gridss-assemble.nix
@@ -57,5 +57,5 @@ stdenv.mkDerivation rec {
       WORKING_DIR=$TMPDIR/ \
       TMP_DIR=$TMPDIR/
   '';
-  passthru.filetype = filetype.bam { ref = ref; sorting = sort.coord {}; };
+  passthru.filetype = filetype.bam { ref = ref; sorting = sort.name {}; };
 }
diff --git a/tools/gridss-softClipsToSplitReads.nix b/tools/gridss-softClipsToSplitReads.nix
index 68ea0a5..c5d8062 100644
--- a/tools/gridss-softClipsToSplitReads.nix
+++ b/tools/gridss-softClipsToSplitReads.nix
@@ -37,5 +37,5 @@ stdenv.mkDerivation rec {
       ${optionalString (config != null) ("OPTIONS_FILE=" + bionix.gridss.gridssConfig config)} \
 			WORKER_THREADS=$NIX_BUILD_CORES
     '';
-  passthru.filetype = filetype.bam { ref = ref; sorting = matchFileSorting "grids-softClipsToSplitReads" { coord = _: input.sorting; name = _: sort.none {}; none = _: input.sorting;} input;};
+  passthru.filetype = filetype.bam { ref = ref; sorting = sort.none {}; };
 }
diff --git a/tools/samtools-sort.nix b/tools/samtools-sort.nix
index e77f3db..37df5c8 100644
--- a/tools/samtools-sort.nix
+++ b/tools/samtools-sort.nix
@@ -25,7 +25,7 @@ in stdenv.mkDerivation {
   buildInputs = [ samtools ];
   buildCommand =
     if alreadySorted then
-      "ln -s $out ${input}"
+      "ln -s ${input} $out"
     else
       ''
         samtools sort -@ $NIX_BUILD_CORES \