update R to 3.5.1 and update full-outer-join to accept multiple table…

… 2's
mskcc · Aug 31, 2021 · 3f1af3e · 3f1af3e
1 parent 712b694
commit 3f1af3e
Show file tree

Hide file tree

Showing 5 changed files with 73 additions and 9 deletions.
diff --git a/Makefile b/Makefile
@@ -177,8 +177,8 @@ singularity-pull-cmoutils:
 	. "$(ENVSH)" singularity && \
 	singularity pull --force --name "$(CMOUTILS_SIF)" docker://$(CMOUTILS_DOCKERTAG)
 
-R_DOCKERTAG:=mskcc/helix_filters_01:R-3.6.1
-R_SIF:=mskcc_helix_filters_01:R-3.6.1.sif
+R_DOCKERTAG:=mskcc/helix_filters_01:R-3.5.1
+R_SIF:=mskcc_helix_filters_01:R-3.5.1.sif
 singularity-pull-r:
 	. "$(ENVSH)" singularity && \
 	singularity pull --force --name "$(R_SIF)" docker://$(R_DOCKERTAG)
@@ -199,7 +199,7 @@ singularity-pull-vep:
 	fi
 # rsync -vrthP "$(VEP_SIF_LOCAL)" "$(VEP_SIF)"
 
-singularity-pull-all: singularity-pull singularity-pull-dev singularity-pull-facets-suite singularity-pull-facets singularity-pull-fillout singularity-pull-igv-reports singularity-pull-cmoutils singularity-pull-msi singularity-pull-vep
+singularity-pull-all: singularity-pull singularity-pull-dev singularity-pull-facets-suite singularity-pull-facets singularity-pull-fillout singularity-pull-igv-reports singularity-pull-cmoutils singularity-pull-msi singularity-pull-vep singularity-pull-r
 
 # change the Docker tag for all the CWL files from the old pattern to the new pattern
 OLD_TAG:=

diff --git a/cwl/full-outer-join.cwl b/cwl/full-outer-join.cwl
@@ -12,7 +12,9 @@ inputs:
     inputBinding:
       position: 1
   table2:
-    type: ['null', File]
+    type:
+      - 'null'
+      - File[] 
     inputBinding:
       position: 2
       prefix: '--t2'

diff --git a/cwl/portal-workflow.cwl b/cwl/portal-workflow.cwl
@@ -149,11 +149,11 @@ inputs:
     type:
     - "null"
     - File
-  extra_cna_file:
+  extra_cna_files:
     doc: "An extra CNA data file to be merged in with the portal CNA data"
     type:
     - "null"
-    - File
+    - File[]
 
 steps:
   # meta_clinical_sample.txt (cbio_clinical_sample_meta_filename; meta_clinical_sample_file)
@@ -366,7 +366,7 @@ steps:
     run: full-outer-join.cwl
     in:
       table1: clean_cna_headers/output_file
-      table2: extra_cna_file
+      table2: extra_cna_files
       join_key:
         valueFrom: ${ return "Hugo_Symbol" }
       output_filename:

diff --git a/tests/test_full-outer-join_cwl.py b/tests/test_full-outer-join_cwl.py
@@ -15,6 +15,68 @@
 class TestMergeTables(PlutoTestCase):
     cwl_file = CWLFile('full-outer-join.cwl')
 
+    def test_merge_three_tables(self):
+        """
+        Test case for merging with three or more tables
+        """
+        lines1 = [
+        ['Hugo_Symbol', 'Sample1', 'Sample2'],
+        ["TAP1", "0", "0"],
+        ["ERRFI1", "0", "0"],
+        ["STK19", "", "0"],
+        ]
+
+        lines2 = [
+        ['Hugo_Symbol', 'Sample3', 'Sample4'],
+        ["ERRFI1", "0", "0"],
+        ["STK19", "-2", "0"],
+        ["STK11", "0", ""],
+        ]
+
+        lines3 = [
+        ['Hugo_Symbol', 'Sample5', 'Sample6'],
+        ["ERRFI1", "-1", ""],
+        ["STK19", "-1", "-2"],
+        ["STK11", "", "-1"],
+        ]
+
+        cna_file1 = self.write_table(self.tmpdir, filename = "cna1.txt", lines = lines1)
+        cna_file2 = self.write_table(self.tmpdir, filename = "cna2.txt", lines = lines2)
+        cna_file3 = self.write_table(self.tmpdir, filename = "cna3.txt", lines = lines3)
+        output_file = os.path.join(self.tmpdir, "output.tsv")
+
+        self.input = {
+            'table1': { "class": "File", "path": cna_file1 },
+            'table2': [{ "class": "File", "path": cna_file2 }, { "class": "File", "path": cna_file3 }, ],
+            'join_key': 'Hugo_Symbol',
+            'output_filename': 'output.tsv'
+        }
+
+        output_json, output_dir = self.run_cwl()
+
+        expected_output = {
+            'output_file': {
+                'location': 'file://' + os.path.join(output_dir,'output.tsv'),
+                'basename': 'output.tsv',
+                'class': 'File',
+                'checksum': 'sha1$9be315be7f1f256d80c1d4e6f9cd31f82381e2c4',
+                'size': 147,
+                'path':  os.path.join(output_dir,'output.tsv')
+                }
+            }
+        self.assertDictEqual(output_json, expected_output)
+
+        output_file = expected_output['output_file']['path']
+        lines = self.read_table(output_file)
+        expected_lines = [
+        ['Hugo_Symbol', 'Sample1', 'Sample2', 'Sample3', 'Sample4', 'Sample5', 'Sample6'],
+        ['ERRFI1', '0', '0', '0', '0', '-1', 'NA'],
+        ['STK19', 'NA', '0', '-2', '0', '-1', '-2'],
+        ['TAP1', '0', '0', 'NA', 'NA', 'NA', 'NA'],
+        ['STK11', 'NA', 'NA', '0', 'NA', 'NA', '-1']
+            ]
+        self.assertEqual(lines, expected_lines)
+
     def test_merge_two_tables(self):
         lines1 = [
         ['Hugo_Symbol', 'Sample1', 'Sample2'],
@@ -33,7 +95,7 @@ def test_merge_two_tables(self):
         cna_file2 = self.write_table(self.tmpdir, filename = "cna2.txt", lines = lines2)
         self.input = {
             'table1': { "class": "File", "path": cna_file1 },
-            'table2': { "class": "File", "path": cna_file2 },
+            'table2': [{ "class": "File", "path": cna_file2 }],
             'join_key': 'Hugo_Symbol',
             'output_filename': 'output.tsv'
         }

diff --git a/tests/test_portal-workflow_cwl.py b/tests/test_portal-workflow_cwl.py
@@ -933,7 +933,7 @@ def test_with_facets_txt_and_facets_mafs(self):
                     "class": "File"
                 },
             ],
-            "extra_cna_file": {"class": "File", "path": cna_file1}
+            "extra_cna_files": [{"class": "File", "path": cna_file1}]
         }
 
         with TemporaryDirectory() as tmpdir: