Merge pull request #34 from cov-lineages/dev

small bug fixes
cov-lineages · Oct 27, 2021 · c0db12f · c0db12f
2 parents e12eb0e + 00a5e56
commit c0db12f
Show file tree

Hide file tree

Showing 4 changed files with 21 additions and 19 deletions.
diff --git a/scorpio/__init__.py b/scorpio/__init__.py
@@ -1,2 +1,2 @@
 _program = "scorpio"
-__version__ = "0.3.13"
+__version__ = "0.3.14"
diff --git a/scorpio/__main__.py b/scorpio/__main__.py
@@ -251,7 +251,7 @@ def main(sysargs = sys.argv[1:]):
         if not args.reference_json:
             args.reference_json = reference_json
             logging.info("Found reference %s" %args.reference_json)
-        if not args.constellations and args.command in ['haplotype', 'classify']:
+        if not args.constellations and args.command in ['haplotype', 'classify', 'list']:
             args.constellations = list_constellation_files
             logging.info("Found constellations:")
             for c in args.constellations:

diff --git a/scorpio/scripts/extract_definitions.py b/scorpio/scripts/extract_definitions.py
@@ -66,7 +66,7 @@ def get_group_dict(in_variants, group_column, index_column, subset):
                 if subset and row[group_column] not in subset:
                     continue
                 if row[index_column] in group_dict:
-                    logging.warning("%s is a duplicate in group CSV, keeping first")
+                    logging.warning("%s is a duplicate in group CSV, keeping first" % row[index_column])
                 else:
                     group_dict[row[index_column]] = row[group_column]
                     groups.add(row[group_column])
@@ -208,17 +208,17 @@ def define_mutations(list_variants, feature_dict, reference_seq, include_protein
                 current[3] = new[3]
         elif current[0] != "":
             var = translate_if_possible(current[1], current[0], current[2], feature_dict, reference_seq, include_protein)
-            if freq:
-                merged_list.append("%s:%s" % (var, freq))
+            if current[3]:
+                merged_list.append("%s:%s" % (var, current[3]))
             else:
                 merged_list.append(var)
             current = new
         else:
             current = new
     if current[0] != "":
         var = translate_if_possible(current[1], current[0], current[2], feature_dict, reference_seq, include_protein)
-        if freq:
-            merged_list.append("%s:%s" % (var, freq))
+        if current[3]:
+            merged_list.append("%s:%s" % (var, current[3]))
         else:
             merged_list.append(var)
     return merged_list
@@ -237,7 +237,7 @@ def subtract_outgroup(common, outgroup_common):
 
 def write_constellation(prefix, group, list_variants, list_intermediates, list_ancestral):
     group_dict = {"name": group, "sites": list_variants, "intermediate": list_intermediates,
-                  "rules": {"min_alt": max(len(list_variants) - 3, min(len(list_variants), 3)), "max_ref": 3}}
+                  "rules": {"min_alt": max(len(list_variants) - 8, min(len(list_variants), 3)), "max_ref": 3}}
     if list_ancestral:
         group_dict["ancestral"] = list_ancestral
     with open('%s/%s.json' % (prefix, group), 'w') as outfile:

diff --git a/scorpio/scripts/type_constellations.py b/scorpio/scripts/type_constellations.py
@@ -8,6 +8,7 @@
 import json
 import re
 import logging
+import copy
 import math
 
 if sys.version_info[0] < 3:
@@ -293,8 +294,8 @@ def parse_json_in(refseq, features_dict, variants_file, constellation_names=None
 
     in_json.close()
     sorted_variants = sorted(variant_list, key=lambda x: int(x["ref_start"]))
-    for var in sorted_variants:
-        print(var)
+    #for var in sorted_variants:
+    #    print(var)
 
     return sorted_variants, name, output_name, rules, mrca_lineage, incompatible_lineage_calls, parent_lineage, lineage_name
 
@@ -868,7 +869,7 @@ def type_constellations(in_fasta, list_constellation_files, constellation_names,
 
             out_list = [record.id]
             for constellation in constellation_dict:
-                barcode_list, counts, constellation_count_dict, sorted_alt_sites = generate_barcode(record.seq, constellation_dict[constellation], ref_char, constellation_count_dict=constellation_count_dict)
+                barcode_list, counts, sample_constellation_count_dict, sorted_alt_sites = generate_barcode(record.seq, constellation_dict[constellation], ref_char, constellation_count_dict=copy.deepcopy(constellation_count_dict))
                 if output_counts or append_genotypes:
                     columns = [record.id]
                     if output_counts:
@@ -878,15 +879,15 @@ def type_constellations(in_fasta, list_constellation_files, constellation_names,
                         columns.extend(barcode_list)
                     if combination:
                         scores = {}
-                        for candidate in constellation_count_dict:
-                            if constellation_count_dict[candidate]["alt"] > 0:
+                        for candidate in sample_constellation_count_dict:
+                            if sample_constellation_count_dict[candidate]["alt"] > 0:
                                 summary = "%s:%i|%i|%i|%i" % (
-                                candidate, constellation_count_dict[candidate]["ref"],
-                                constellation_count_dict[candidate]["alt"],
-                                constellation_count_dict[candidate]["ambig"],
-                                constellation_count_dict[candidate]["oth"])
-                                score = float(constellation_count_dict[candidate]["alt"]) / \
-                                        constellation_count_dict[candidate]["total"]
+                                candidate, sample_constellation_count_dict[candidate]["ref"],
+                                sample_constellation_count_dict[candidate]["alt"],
+                                sample_constellation_count_dict[candidate]["ambig"],
+                                sample_constellation_count_dict[candidate]["oth"])
+                                score = float(sample_constellation_count_dict[candidate]["alt"]) / \
+                                        sample_constellation_count_dict[candidate]["total"]
                                 scores[score] = summary
                         sorted_scores = sorted(scores, key=lambda x: float(x), reverse=True)
                         columns.append("; ".join([scores[score] for score in sorted_scores]))
@@ -1086,6 +1087,7 @@ def list_constellations(list_constellation_files, constellation_names, reference
             list_of_constellations.add(mrca_lineage)
         elif len(variants) > 0:
             list_of_constellations.add(output_name)
+    print("\n".join(list_of_constellations))
 
 def parse_args():
     parser = argparse.ArgumentParser(description="""Type an alignment at specific sites and classify with a barcode.""",