diff --git a/query_ror.py b/query_ror.py index fbe5e3e..6a32631 100644 --- a/query_ror.py +++ b/query_ror.py @@ -503,6 +503,7 @@ def site_sponsor(args): sponsor_country_column = args.sponsor_country_column site_country_column = args.site_country_column exclude_indiv_company = args.exclude_indiv_company + exclude_same = args.exclude_same site_df = load_glob(site_files, site_filter, exclude_indiv_company) sponsor_df = load_glob(sponsor_files, sponsor_filter, exclude_indiv_company) @@ -521,6 +522,9 @@ def site_sponsor(args): .trial_id.count() .reset_index() ) + + if exclude_same: + counts = counts.loc[counts.who_region != counts.sponsor_who_region] # Map nodes to node ids who_map = {name: index for index, name in enumerate(counts.who_region.unique())} who_sponsor_map = { @@ -788,6 +792,11 @@ def multisite(args): action="store_true", help="Exclude individuals and companies", ) + site_sponsor_parser.add_argument( + "--exclude-same", + action="store_true", + help="Exclude site/sponsor region the same", + ) site_sponsor_parser.set_defaults(func=site_sponsor) args = ror_parser.parse_args()