pass update prep metrics and payload modules

Author: guanqiaofeng

modules.json

@@ -41,8 +41,8 @@
                         ]
                     },
                     "prep/metrics": {
-                        "branch": "main",
-                        "git_sha": "c041bcd0e203622a9ebebd720897d162a9ba95d7",
+                        "branch": "update_payload_2",
+                        "git_sha": "840aac9dffe16e09013b263b6982777d72860192",
                         "installed_by": [
                             "modules"
                         ]

modules/icgc-argo-workflows/prep/metrics/main.nf

@@ -21,8 +21,8 @@ process PREP_METRICS {
 
     script:
     def workflow_name = workflow.Manifest.name
-
-
+    
+    
     """
     case '$workflow_name' in
     'Pre Alignment QC')

modules/icgc-argo-workflows/prep/metrics/resources/usr/bin/dnaaln.py

@@ -27,6 +27,7 @@
 import csv
 
 ga4gh_wgs_qc_metrics = [
+ 
    'insert_size_std_deviation',
    'mad_autosome_coverage',
    'mean_autosome_coverage',
@@ -79,7 +80,7 @@ def get_mqc_stats(multiqc, sampleId):
     for f in sorted(glob(multiqc+'/*.txt')):
       for tool_metrics in tool_fieldmap.keys():
         if f.endswith(tool_metrics+'.txt'):
-          with open(f, 'r') as fn:
+          with open(f, 'r') as fn: 
             mqc_stats[tool_metrics] = []
             reader = csv.DictReader(fn, delimiter="\t")
             for row in reader:
@@ -105,7 +106,7 @@ def get_mqc_stats(multiqc, sampleId):
         mqc_stats['metrics'].update({
           fn: new_value
         })
-
+    
     return mqc_stats
 
 def main():
@@ -119,7 +120,7 @@ def main():
     parser.add_argument("-q", "--qc_files", dest="qc_files", required=True, type=str, nargs="+", help="qc files")
 
     args = parser.parse_args()
-
+    
     # get tool_specific & aggregated metrics from multiqc
     mqc_stats = {}
     if args.multiqc:
@@ -132,7 +133,7 @@ def main():
         file_type = 'gatk_contamination'
         if file_type in mqc_stats: continue
         mqc_stats[file_type] = []
-        with open(fn, 'r') as f:
+        with open(fn, 'r') as f:     
           reader = csv.DictReader(f, delimiter="\t")
           for row in reader:
             mqc_stats[file_type].append(row)
@@ -155,7 +156,7 @@ def main():
     }
     for k,v in mqc_stats_updated.get('metrics', None).items():
         if not k in ga4gh_wgs_qc_metrics: continue
-        ga4gh_qc_dict['wgs_qc_metrics'].update({k: v})
+        ga4gh_qc_dict['wgs_qc_metrics'].update({k: v}) 
 
     if ga4gh_qc_dict['wgs_qc_metrics']:
       with open("%s.dnaaln.metrics.json" % (args.sampleId), 'w') as f:

modules/icgc-argo-workflows/prep/metrics/resources/usr/bin/rnaaln.py

@@ -30,7 +30,6 @@
     'hisat2': {
         'paired_total' : 'paired_total',
         'unpaired_total' : 'unpaired_total'
-        # 'overall_alignment_rate' : 'overall_alignment_rate'  # collected from samtools stats
     },
     'star' : {
         'total_reads' : 'total_reads',
@@ -39,7 +38,6 @@
         'avg_mapped_read_length' : 'avg_mapped_read_length',
         'num_splices' : 'num_splices',
         'num_annotated_splices' : 'num_annotated_splices',
-        # 'mismatch_rate' : 'mismatch_rate', # collected from samtools
         'pct_multimapped' : 'multimapped_percent'
     },
     'picard_RnaSeqMetrics': {
@@ -81,7 +79,7 @@ def get_mqc_stats(multiqc, sampleId):
     for f in sorted(glob(multiqc+'/*.txt')):
         for tool_metrics in tool_fieldmap.keys():
             if f.endswith(tool_metrics+'.txt'):
-                with open(f, 'r') as fn:
+                with open(f, 'r') as fn: 
                     mqc_stats[tool_metrics] = []
                     reader = csv.DictReader(fn, delimiter="\t")
                     for row in reader:
@@ -94,7 +92,7 @@ def get_mqc_stats(multiqc, sampleId):
 
     # convert the fraction to percentage for given fields
     for fn in fra2pct_fields:
-        if not mqc_stats['metrics'].get(fn):
+        if not mqc_stats['metrics'].get(fn): 
             print(f"Field '{fn}' not found in mqc_stats['metrics'] dictionary")
             continue
         new_value = round(float(mqc_stats['metrics'][fn]) * 100, 2)