From 9cf1724895f001c625fba37d78f517acb6938e07 Mon Sep 17 00:00:00 2001
From: Justin Bedo <cu@cua0.org>
Date: Wed, 2 Sep 2020 16:24:37 +1000
Subject: gridss: shard assembly into 10 pieces by default

---
 tools/gridss-assemble.nix | 15 +++++++++++++--
 tools/gridss-variants.nix |  3 ++-
 tools/gridss.nix          |  3 +++
 3 files changed, 18 insertions(+), 3 deletions(-)

(limited to 'tools')

diff --git a/tools/gridss-assemble.nix b/tools/gridss-assemble.nix
index a532f4b..f0a2693 100644
--- a/tools/gridss-assemble.nix
+++ b/tools/gridss-assemble.nix
@@ -6,6 +6,9 @@
 , flags ? null
 , config ? null
 , heapSize ? "31g"
+, workdirs ? []
+, jobIndex ? null
+, jobNodes ? null
 }:
 
 with bionix;
@@ -37,7 +40,8 @@ assert (homoRef);
 
 stage rec {
   name = "gridss-assemble";
-  buildInputs = with pkgs; [ jre bwa ];
+  buildInputs = with pkgs; [ jre bwa rsync ];
+  outputs = [ "out" "work" ];
   buildCommand = ''
     TMPDIR=$(pwd)
     ln -s ${ref} ref.fa
@@ -46,7 +50,8 @@ stage rec {
       ln -s $f
     done
     ${concatMapStringsSep "\n" linkInput inputs}
-	  java -Xmx${heapSize} -Dsamjdk.create_index=true \
+    ${concatMapStringsSep "\n" (w: "rsync -a --ignore-existing ${w}/ ./") workdirs}
+    java -Xmx${heapSize} -Dsamjdk.create_index=true \
       -cp ${bionix.gridss.jar} gridss.AssembleBreakends \
       VERBOSITY=WARNING \
       REFERENCE_SEQUENCE=ref.fa \
@@ -56,7 +61,13 @@ stage rec {
       ${optionalString (config != null) ("OPTIONS_FILE=" + bionix.gridss.gridssConfig config)} \
       WORKING_DIR=$TMPDIR/ \
       TMP_DIR=$TMPDIR/ \
+      ${optionalString (jobIndex != null) "JOB_INDEX=${toString jobIndex}"} \
+      ${optionalString (jobIndex != null) "JOB_NODES=${toString jobNodes}"} \
       ${optionalString (flags != null) flags}
+    rm -rf tmp
+    touch $out
+    cp -r $TMPDIR $work
+    chmod u+rwX -R $work
   '';
   passthru.filetype = filetype.bam { ref = ref; sorting = sort.name {}; };
   passthru.multicore = true;
diff --git a/tools/gridss-variants.nix b/tools/gridss-variants.nix
index ad4c66d..c208758 100644
--- a/tools/gridss-variants.nix
+++ b/tools/gridss-variants.nix
@@ -7,6 +7,7 @@
 , softClipsToSplitReadsAttrs ? { flags = "REALIGN_ENTIRE_READ=true"; }
 , config ? null
 , heapSize ? "4g"
+, shards ? 10
 }:
 
 with bionix;
@@ -43,7 +44,7 @@ let
     ln -s ${bionix.samtools.index indexAttrs input} $WRKDIR/$BASENAME.sv.bai
   '';
 
-  assembly = bionix.samtools.sort {} (softClipsToSplitReads softClipsToSplitReadsAttrs (bionix.gridss.assemble assemblyAttrs inputs));
+  assembly = bionix.samtools.sort {} (softClipsToSplitReads softClipsToSplitReadsAttrs (bionix.gridss.shardedAssemble shards assemblyAttrs inputs));
 
   mkLinks = ''
     ln -s ${ref} ref.fa
diff --git a/tools/gridss.nix b/tools/gridss.nix
index aaf8e16..41f6a5a 100644
--- a/tools/gridss.nix
+++ b/tools/gridss.nix
@@ -43,6 +43,9 @@ rec {
   Type: assemble :: {config :: ini = null, heapSize :: String = "31g", ...} -> [bam] -> bam
   */
   assemble = callBionixE ./gridss-assemble.nix;
+  shardedAssemble = n: a: input:
+    let assemblies = genList (i: bionix.gridss.assemble (a // { jobNodes = n; jobIndex = i;}) input) n;
+    in if n <= 1 then bionix.gridss.assemble a input else bionix.gridss.assemble (a // {workdirs = map (a: a.work) assemblies;}) input;
 
   /* Invoke identifyVariants tool
   Type: identifyVariants :: {config :: ini = null, heapSize :: String = "4g", ...} -> [bam] -> VCF
-- 
cgit v1.2.3