Merge pull request #588 from 4dn-dcic/chip_duplicate_ctls

Deduplicate control TAs in ChIP-seq input JSON

CHANGELOG.rst

@@ -7,6 +7,14 @@ foursight
 Change Log
 ----------
 
+4.9.10
+=====
+
+`PR 588: Deduplicate control TAs in ChIP-seq input JSON <https://github.com/4dn-dcic/foursight/pull/588>`_
+
+* For ChIP-seq sets with 2+ identical control TA files, replace multi-file ctl list with one-item list
+
+
 4.9.9
 =====
 

chalicelib_fourfront/checks/wfr_encode_checks.py

@@ -388,6 +388,10 @@ def rename_chip(input_at_id_list):
                 if ta_cnt:
                     s2_input_files['chip.ctl_tas'] = ta_cnt
                     s2_input_files['additional_file_parameters']['chip.ctl_tas'] = {"rename": rename_chip(ta_cnt)}
+                    if len(set(ta_cnt)) == 1 and len(ta_cnt) != 1:
+                        print("Control TAs are identical, listing file only once")
+                        s2_input_files['chip.ctl_tas'] = [ta_cnt[0]]
+                        s2_input_files['additional_file_parameters']['chip.ctl_tas'] = {"rename": rename_chip([ta_cnt[0]])}
 
                 # collect parameters
                 parameters = {}

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "foursight"
-version = "4.9.9"
+version = "4.9.10"
 description = "Serverless Chalice Application for Monitoring"
 authors = ["4DN-DCIC Team <[email protected]>"]
 license = "MIT"