From 535b8017ddd27a9db683f6d29368258b5c48cf5a Mon Sep 17 00:00:00 2001
From: Peter Amstutz
Date: Fri, 3 Jul 2020 20:45:17 +0000
Subject: Improving genbank import workflow

---
 scripts/split_into_arrays.cwl | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)
 create mode 100644 scripts/split_into_arrays.cwl

(limited to 'scripts/split_into_arrays.cwl')

diff --git a/scripts/split_into_arrays.cwl b/scripts/split_into_arrays.cwl
new file mode 100644
index 0000000..102fe7d
--- /dev/null
+++ b/scripts/split_into_arrays.cwl
@@ -0,0 +1,30 @@
+cwlVersion: v1.1
+class: ExpressionTool
+requirements:
+  InlineJavascriptRequirement: {}
+inputs:
+  dir:
+    type: Directory
+    loadListing: shallow_listing
+outputs:
+  fasta: File[]
+  metadata: File[]
+expression: |
+  ${
+  var dir = inputs.dir;
+  var fasta = [];
+  var metadata = [];
+  dir.listing.sort(function(a, b) { return a.basename < b.basename; });
+  for (var i = 0; i < dir.listing.length; i++) {
+    if (dir.listing[i].basename.substr(-6) == ".fasta") {
+      fasta.push(dir.listing[i]);
+    }
+    if (dir.listing[i].basename.substr(-5) == ".yaml") {
+      metadata.push(dir.listing[i]);
+    }
+  }
+  if (fasta.length != metadata.length) {
+    throw "They dont match";
+  }
+  return {"fasta": fasta, "metadata": metadata};
+  }
-- 
cgit v1.2.3