Workaround CWL limit by chunking file list

Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>

4 files changed, 79 insertions, 30 deletions

diff --git a/workflows/pangenome-generate/merge-metadata.cwl b/workflows/pangenome-generate/merge-metadata.cwl
index fcefe32..4d9c808 100644
--- a/workflows/pangenome-generate/merge-metadata.cwl
+++ b/workflows/pangenome-generate/merge-metadata.cwl
@@ -5,16 +5,48 @@ hints:
     dockerPull: commonworkflowlanguage/cwltool_module
 inputs:
   metadata: File[]
-  metadataSchema: File
   subjects: string[]
-  dups: File?
-  originalLabels: File
+  metadataSchema:
+    type: File
+    inputBinding: {position: 2}
+  originalLabels:
+    type: File
+    inputBinding: {position: 3}
+  dups:
+    type: File?
+    inputBinding: {position: 4}
+  script:
+    type: File
+    inputBinding: {position: 1}
+    default: {class: File, location: merge-metadata.py}
 outputs:
   merged: stdout
 stdout: mergedmetadata.ttl
 requirements:
+  InlineJavascriptRequirement: {}
   InitialWorkDirRequirement:
-    listing:
-      - entry: {$include: merge-metadata.py}
-        entryname: merge-metadata.py
-baseCommand: [python3, merge-metadata.py]
+    listing: |
+          ${
+          var i = 0;
+          var b = 1;
+          var out = [];
+          for (; i < inputs.metadata.length; i++) {
+            var block = [];
+            var sub = [];
+            for (; i < (b*150) && i < inputs.metadata.length; i++) {
+              block.push(inputs.metadata[i]);
+              sub.push(inputs.subjects[i]);
+            }
+            out.push({
+              entryname: "block"+b,
+              entry: JSON.stringify(block)
+            });
+            out.push({
+              entryname: "subs"+b,
+              entry: JSON.stringify(sub)
+            });
+            b++;
+          }
+          return out;
+          }
+baseCommand: python
diff --git a/workflows/pangenome-generate/merge-metadata.py b/workflows/pangenome-generate/merge-metadata.py
index bfec781..65d08a6 100644
--- a/workflows/pangenome-generate/merge-metadata.py
+++ b/workflows/pangenome-generate/merge-metadata.py
@@ -2,12 +2,27 @@ import re
 import schema_salad.schema
 import schema_salad.jsonld_context
 import json
+import sys
+import os
+import logging
+
+metadataSchema = sys.argv[1]
+originalLabels = sys.argv[2]
+dups = None
+if len(sys.argv) == 4:
+    dups = sys.argv[3]
+
+def readitems(stem):
+    items = []
+    b = 1
+    while os.path.exists("%s%i" % (stem, b)):
+        with open("%s%i" % (stem, b)) as f:
+            items.extend(json.load(f))
+        b += 1
+    return items
 
-metadataSchema = '$(inputs.metadataSchema.path)'
-metadata = $(inputs.metadata)
-subjects = $(inputs.subjects)
-dups = json.loads('''$(inputs.dups)''')
-originalLabels = $(inputs.originalLabels)
+metadata = readitems("block")
+subjects = readitems("subs")
 
 (document_loader,
  avsc_names,
@@ -20,17 +35,15 @@ for i, m in enumerate(metadata):
     g = schema_salad.jsonld_context.makerdf(subjects[i], doc, document_loader.ctx)
     print(g.serialize(format="ntriples").decode("utf-8"))
 
-import logging
-
 if dups:
-    sameseqs = open(dups["path"], "rt")
+    sameseqs = open(dups, "rt")
     for d in sameseqs:
         logging.warn(d)
-        g = re.match(r"\\d+\\t(.*)", d)
+        g = re.match(r"\d+\t(.*)", d)
         logging.warn("%s", g.group(1))
         sp = g.group(1).split(",")
         for n in sp[1:]:
             print("<%s> <http://biohackathon.org/bh20-seq-schema/has_duplicate_sequence> <%s> ." % (n.strip(), sp[0].strip()))
 
-orig = open(originalLabels["path"], "rt")
+orig = open(originalLabels, "rt")
 print(orig.read())
diff --git a/workflows/pangenome-generate/relabel-seqs.cwl b/workflows/pangenome-generate/relabel-seqs.cwl
index 01196f6..c1f17a4 100644
--- a/workflows/pangenome-generate/relabel-seqs.cwl
+++ b/workflows/pangenome-generate/relabel-seqs.cwl
@@ -26,19 +26,21 @@ requirements:
           var out = [];
           for (; i < inputs.readsFA.length; i++) {
             var block = [];
-            for (; i < (b*100) && i < inputs.readsFA.length; i++) {
+            var sub = [];
+            for (; i < (b*150) && i < inputs.readsFA.length; i++) {
               block.push(inputs.readsFA[i]);
+              sub.push(inputs.subjects[i]);
             }
             out.push({
               entryname: "block"+b,
               entry: JSON.stringify(block)
             });
+            out.push({
+              entryname: "subs"+b,
+              entry: JSON.stringify(sub)
+            });
             b++;
           }
-          out.push({
-            entry: JSON.stringify(inputs.subjects),
-            entryname: "subjects"
-          });
           return out;
           }
 hints:
diff --git a/workflows/pangenome-generate/relabel-seqs.py b/workflows/pangenome-generate/relabel-seqs.py
index 970540f..6b022a0 100644
--- a/workflows/pangenome-generate/relabel-seqs.py
+++ b/workflows/pangenome-generate/relabel-seqs.py
@@ -1,15 +1,17 @@
 import os
 import json
 
-reads = []
-b = 1
-while os.path.exists("block%i" % b):
-    with open("block%i" % b) as f:
-        reads.extend(json.load(f))
-    b += 1
+def readitems(stem):
+    items = []
+    b = 1
+    while os.path.exists("%s%i" % (stem, b)):
+        with open("%s%i" % (stem, b)) as f:
+            items.extend(json.load(f))
+        b += 1
+    return items
 
-with open("subjects") as f:
-    subjects = json.load(f)
+reads = readitems("block")
+subjects = readitems("subs")
 
 relabeled_fasta = open("relabeledSeqs.fasta", "wt")
 original_labels = open("originalLabels.ttl", "wt")