Merge pull request #551 from 4dn-dcic/repliseq_v16.1

Upgrade foursight to run repliseq v16.1

CHANGELOG.rst

@@ -7,6 +7,13 @@ foursight
 Change Log
 ----------
 
+4.0.3
+=====
+
+`PR 551: Upgrade foursight to run repliseq v16.1 <https://github.com/4dn-dcic/foursight/pull/551>`_
+
+* configure Repli-seq pipeline to run v16.1, introducing additional output file
+
 4.0.2
 =====
 

chalicelib_fourfront/checks/helpers/wfr_utils.py

@@ -59,7 +59,7 @@
     },
     "repliseq-parta": {
         "run_time": 200,
-        "accepted_versions": ["v13.1", "v14", "v16"]
+        "accepted_versions": ["v13.1", "v14", "v16","v16.1"]
     },
     "bedGraphToBigWig": {
         "run_time": 24,
@@ -185,17 +185,20 @@
     # Preliminary -  Don't release - (Released to network is pending approval from Belmont lab)
     'TSA-seq':       ['RepliSeq_Pipeline_v13.1_step1',
                       'RepliSeq_Pipeline_v14_step1',
-                      'RepliSeq_Pipeline_v16_step1'],
+                      'RepliSeq_Pipeline_v16_step1',
+                      'RepliSeq_Pipeline_v16.1_step1'],
     # OFFICIAL - 1 STEP
     '2-stage Repli-seq': ['RepliSeq_Pipeline_v13.1_step1',
                           'RepliSeq_Pipeline_v14_step1',
-                          'RepliSeq_Pipeline_v16_step1'],
+                          'RepliSeq_Pipeline_v16_step1',
+                          'RepliSeq_Pipeline_v16.1_step1'],
     # OFFICIAL - 1 STEP
     'Multi-stage Repli-seq': ['RepliSeq_Pipeline_v13.1_step1',
                               'RepliSeq_Pipeline_v14_step1',
-                              'RepliSeq_Pipeline_v16_step1'],
+                              'RepliSeq_Pipeline_v16_step1',
+                              'RepliSeq_Pipeline_v16.1_step1'],
     # Preliminary - Released to network
-    'NAD-seq':       ['RepliSeq_Pipeline_v13.1_step1', 'RepliSeq_Pipeline_v14_step1', 'RepliSeq_Pipeline_v16_step1'],
+    'NAD-seq':       ['RepliSeq_Pipeline_v13.1_step1', 'RepliSeq_Pipeline_v14_step1', 'RepliSeq_Pipeline_v16_step1', 'RepliSeq_Pipeline_v16.1_step1'],
     # OFFICIAL
     'ATAC-seq':      ['ENCODE_ATAC_Pipeline_1.1.1'],
     # OFFICIAL
@@ -1702,6 +1705,7 @@ def check_repli(res, my_auth, tag, check, start, lambda_limit, winsize=None):
                 # if successful
                 if step1_result['status'] == 'complete':
                     all_files.extend([step1_result['filtered_sorted_deduped_bam'],
+                                      step1_result['count_bg_rpkm'],
                                       step1_result['count_bg']])
                 # if still running
                 elif step1_result['status'] == 'running':

chalicelib_fourfront/checks/helpers/wfrset_utils.py

@@ -150,13 +150,17 @@ def step_settings(step_name, my_organism, attribution, overwrite=None):
     },
     {
         'app_name': 'repliseq-parta',
-        'workflow_uuid': '4dn-dcic-lab:wf-repliseq-parta-v16',
+        'workflow_uuid': '4dn-dcic-lab:wf-repliseq-parta-v16.1',
         "parameters": {"nthreads": 4, "memperthread": "2G"},
         'custom_pf_fields': {
             'filtered_sorted_deduped_bam': {
                 'genome_assembly': genome,
                 'file_type': 'alignments',
                 'description': 'This is an output file of the RepliSeq processing pipeline'},
+            'count_bg_rpkm': {
+                'genome_assembly': genome,
+                'file_type': 'normalized counts',
+                'description': 'read counts, unfiltered RPKM'},
             'count_bg': {
                 'genome_assembly': genome,
                 'file_type': 'counts',

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "foursight"
-version = "4.0.2"
+version = "4.0.3"
 description = "Serverless Chalice Application for Monitoring"
 authors = ["4DN-DCIC Team <[email protected]>"]
 license = "MIT"