octopus: initial functions

5 files changed, 103 insertions, 0 deletions

diff --git a/default.nix b/default.nix
index 9bc7b6c..3077470 100644
--- a/default.nix
+++ b/default.nix
@@ -36,6 +36,7 @@ let
     strelka = callBionix ./tools/strelka.nix {};
     ascat = callBionix ./tools/ascat.nix {};
     fastp = callBionix ./tools/fastp.nix {};
+    octopus = callBionix ./tools/octopus.nix {};
 
     slurm = attrs: bionix.extend (self: super: with self; rec {
       slurmDefs = { ppn = 1; mem = 1; walltime = "24:00:00"; partition = null; slurmFlags = null; salloc = "/usr/bin/salloc"; srun = "/usr/bin/srun"; } // attrs;
diff --git a/test-tnpair.nix b/test-tnpair.nix
index 76fcfab..f90e3a9 100644
--- a/test-tnpair.nix
+++ b/test-tnpair.nix
@@ -37,8 +37,10 @@ let
   processPair = { tumour, normal }: rec {
     alignments = mapAttrs (_: x: markdup (sort (fixmate (alignWithRG x.name x.files)))) { inherit normal tumour; };
     variants = callVariants alignments;
+    octopusSomatic = octopus.callSomatic {} {inherit (alignments) normal; tumours = [ alignments.tumour ];};
     glvariants = strelka.call {} (builtins.attrValues alignments);
     platypusVars = platypus.call {} (builtins.attrValues alignments);
+    octopusVars = octopus.call {} (builtins.attrValues alignments);
     shards = pipe [
       (shard.fastQPair 2)
       (map (bwa.align {inherit ref;}))
@@ -56,6 +58,8 @@ let
     (ln (facets.callCNV {} {vcf = tnpairResult.platypusVars; bams = with tnpairResult.alignments; [ normal tumour ];}) "facets")
     (ln cnvkitResults.cnvs "cnvkit")
     (ln cnvkitResults.plot "cnvkit.pdf")
+    (ln tnpairResult.octopusVars "octopus.vcf")
+    (ln tnpairResult.octopusSomatic "octopus-somatic.vcf")
     (ln tnpairResult.variants "strelka")
     (ln tnpairResult.glvariants "strelka-gl")
     (ln tnpairResult.variants.indels "strelka.indels.vcf")
diff --git a/tools/octopus-call.nix b/tools/octopus-call.nix
new file mode 100644
index 0000000..f88960c
--- /dev/null
+++ b/tools/octopus-call.nix
@@ -0,0 +1,37 @@
+{ bionix
+, faidxAttrs ? {}
+, indexAttrs ? {}
+, flags ? ""}:
+
+with bionix;
+with lib;
+with types;
+
+inputs:
+
+let
+  getref = f: matchFiletype "octopus-callSomatic" { bam = {ref, ...}: ref; cram = {ref, ...}: ref;} f;
+  refs = map getref inputs;
+  ref = head refs;
+
+in
+
+assert (length (unique refs) == 1);
+
+stage {
+  name = "octopus-call";
+  buildInputs = with pkgs; [ octopus-caller ];
+  buildCommand = ''
+    ln -s ${ref} ref.fa
+    ln -s ${samtools.faidx faidxAttrs ref} ref.fai
+    ${concatMapStringsSep "\n" (i: ''
+      ln -s ${i} $(basename ${i}).bam
+      ln -s ${samtools.index indexAttrs i} $(basename ${i}).bai
+    '') inputs}
+    octopus -R ref.fa -I *.bam -o $out \
+      --threads=$NIX_BUILD_CORES \
+      ${flags}
+  '';
+  passthru.filtype = filetype.vcf {ref = ref;};
+  passthru.multicore = true;
+}
diff --git a/tools/octopus-callSomatic.nix b/tools/octopus-callSomatic.nix
new file mode 100644
index 0000000..e1c7ce6
--- /dev/null
+++ b/tools/octopus-callSomatic.nix
@@ -0,0 +1,53 @@
+{ bionix
+, faidxAttrs ? {}
+, indexAttrs ? {}
+, flags ? ""}:
+
+with bionix;
+with lib;
+with types;
+
+{normal, tumours}:
+
+let
+  smScript = pkgs.writeText "smScript.awk" ''
+    /^@RG/{
+      for(i = 1; i <= NF; i++) {
+        n=split($i, fields, ":")
+        if(n == 2 && fields[1] == "SM"){
+          print fields[2]
+          exit
+        }
+      }
+    }
+  '';
+
+  inputs = [normal] ++ tumours;
+  getref = f: matchFiletype "octopus-callSomatic" { bam = {ref, ...}: ref; cram = {ref, ...}: ref;} f;
+  refs = map getref inputs;
+  ref = head refs;
+
+in
+
+assert (length (unique refs) == 1);
+
+
+stage {
+  name = "octopus-callSomatic";
+  buildInputs = with pkgs; [ octopus-caller samtools ];
+  buildCommand = ''
+    ln -s ${ref} ref.fa
+    ln -s ${samtools.faidx faidxAttrs ref} ref.fai
+    ${concatMapStringsSep "\n" (i: ''
+      ln -s ${i} $(basename ${i}).bam
+      ln -s ${samtools.index indexAttrs i} $(basename ${i}).bai
+    '') inputs}
+    normal=$(samtools view -H ${normal} | awk -f ${smScript})
+    octopus -R ref.fa -I *.bam -o $out \
+      --threads=$NIX_BUILD_CORES \
+      -N $normal \
+      ${flags}
+  '';
+  passthru.filetype = filetype.vcf {ref = ref;};
+  passthru.multicore = true;
+}
diff --git a/tools/octopus.nix b/tools/octopus.nix
new file mode 100644
index 0000000..4d0c7f6
--- /dev/null
+++ b/tools/octopus.nix
@@ -0,0 +1,8 @@
+{ bionix }:
+
+with bionix;
+
+{
+  call = callBionixE ./octopus-call.nix;
+  callSomatic = callBionixE ./octopus-callSomatic.nix;
+}