Merge pull request #42 from cov-lineages/dev

Handle multiple levels of definition within a constellation

scorpio/__init__.py

@@ -1,2 +1,2 @@
 _program = "scorpio"
-__version__ = "0.3.15"
+__version__ = "0.3.16"

scorpio/scripts/type_constellations.py

@@ -295,7 +295,10 @@ def parse_json_in(refseq, features_dict, variants_file, constellation_names=None
                 variant_list.append(record)
 
     if "rules" in json_dict:
-        rules = json_dict["rules"]
+        if type(json_dict["rules"]) == dict and "default" in json_dict["rules"]:
+            rules = json_dict["rules"]
+        else:
+            rules = {"default": json_dict["rules"]}
 
     in_json.close()
     sorted_variants = sorted(variant_list, key=lambda x: int(x["ref_start"]))
@@ -346,9 +349,9 @@ def parse_csv_in(refseq, features_dict, variants_file, constellation_names=None,
     csv_in.close()
     rules = None
     if len(compulsory) > 0:
-        rules = {}
+        rules = {"default": {}}
         for var in compulsory:
-            rules[var] = "alt"
+            rules["default"][var] = "alt"
     sorted_variants = sorted(variant_list, key=lambda x: int(x["ref_start"]))
     return sorted_variants, name, rules
 
@@ -570,24 +573,24 @@ def var_follows_rules(call, rule):
     else:
         return call == rule_call
 
-def counts_follow_rules(counts, rules):
+def counts_follow_rules(counts, rules, key):
     # rules allowed include "max_ref", "min_alt", "min_snp_alt"
     is_rule_follower = True
     notes = []
-    for rule in rules:
+    for rule in rules[key]:
         if ":" in rule:
             continue
         elif str(rule).startswith("min") or str(rule).startswith("max"):
             rule_parts = rule.split("_")
             if len(rule_parts) <= 1:
                 continue
             elif len(rule_parts) == 2:
-                if rule_parts[0] == "min" and counts[rule_parts[1]] < rules[rule]:
+                if rule_parts[0] == "min" and counts[rule_parts[1]] < rules[key][rule]:
                     is_rule_follower = False
-                elif rule_parts[0] == "max" and counts[rule_parts[1]] > rules[rule]:
+                elif rule_parts[0] == "max" and counts[rule_parts[1]] > rules[key][rule]:
                     is_rule_follower = False
                 else:
-                    counts["rules"] += 1
+                    counts["rules"][key] += 1
             elif len(rule_parts) == 3:
                 part = None
                 if rule_parts[1] in ["substitution", "snp"]:
@@ -596,24 +599,27 @@ def counts_follow_rules(counts, rules):
                     part = "indel"
                 if not part:
                     is_rule_follower = False
-                elif rule_parts[0] == "min" and counts[part][rule_parts[2]] < rules[rule]:
+                elif rule_parts[0] == "min" and counts[part][rule_parts[2]] < rules[key][rule]:
                     is_rule_follower = False
-                    notes.append("%s_%s_count=%i is less than %i" % (part, rule_parts[2], counts[part][rule_parts[2]], rules[rule]))
-                elif rule_parts[0] == "max" and counts[part][rule_parts[2]] > rules[rule]:
+                    notes.append("%s_%s_count=%i is less than %i" % (part, rule_parts[2], counts[part][rule_parts[2]], rules[key][rule]))
+                elif rule_parts[0] == "max" and counts[part][rule_parts[2]] > rules[key][rule]:
                     is_rule_follower = False
-                    notes.append("%s_%s_count=%i is more than %i" % (part, rule_parts[2], counts[part][rule_parts[2]], rules[rule]))
+                    notes.append("%s_%s_count=%i is more than %i" % (part, rule_parts[2], counts[part][rule_parts[2]], rules[key][rule]))
                 else:
-                    counts["rules"] += 1
+                    counts["rules"][key] += 1
         else:
-            logging.warning("Warning: Ignoring rule %s:%s" % (rule, str(rules[rule])))
+            logging.warning("Warning: Ignoring rule %s:%s" % (rule, str(rules[key][rule])))
     return is_rule_follower, ";".join(notes)
 
 def count_and_classify(record_seq, variant_list, rules):
     assert rules is not None
-    counts = {'ref': 0, 'alt': 0, 'ambig': 0, 'oth': 0, 'rules': 0,
+    counts = {'ref': 0, 'alt': 0, 'ambig': 0, 'oth': 0, 'rules': {},
               'substitution': {'ref': 0, 'alt': 0, 'ambig': 0, 'oth': 0},
               'indel': {'ref': 0, 'alt': 0, 'ambig': 0, 'oth': 0}}
-    is_rule_follower = True
+    is_rule_follower_dict = {}
+    for key in rules:
+        is_rule_follower_dict[key] = True
+        counts["rules"][key] = 0
 
     for var in variant_list:
         call, query_allele = call_variant_from_fasta(record_seq, var)
@@ -623,20 +629,27 @@ def count_and_classify(record_seq, variant_list, rules):
             counts["substitution"][call] += 1
         elif var['type'] in ["ins", "del"]:
             counts["indel"][call] += 1
-        if var["name"] in rules:
-            if var_follows_rules(call, rules[var["name"]]):
-                counts['rules'] += 1
-            elif is_rule_follower:
-                is_rule_follower = False
+        for key in rules:
+            if var["name"] in rules[key]:
+                if var_follows_rules(call, rules[key][var["name"]]):
+                    counts['rules'][key] += 1
+                elif is_rule_follower_dict[key]:
+                    is_rule_follower_dict[key] = False
 
     counts['support'] = round(counts['alt']/float(counts['alt'] + counts['ref'] + counts['ambig'] + counts['oth']),4)
     counts['conflict'] = round(counts['ref'] /float(counts['alt'] + counts['ref'] + counts['ambig'] + counts['oth']),4)
 
-    if not is_rule_follower:
-        return counts, False, ""
-    else:
-        call, note = counts_follow_rules(counts, rules)
-        return counts, call, note
+    for key in rules:
+        if not is_rule_follower_dict[key]:
+            continue
+        else:
+            call, note = counts_follow_rules(counts, rules, key)
+            if call:
+                counts["rules"] = counts["rules"][key]
+                call = key
+                return counts, call, note
+    counts["rules"] = counts["rules"]["default"]
+    return counts, False, ""
 
 
 def generate_barcode(record_seq, variant_list, ref_char=None, ins_char="?", oth_char="X",constellation_count_dict=None):
@@ -920,7 +933,15 @@ def combine_counts_call_notes(counts1, call1, note1, counts2, call2, note2):
             counts[key] = counts1[key] + counts2[key]
     counts['support'] = round(counts['alt'] / float(counts['alt'] + counts['ref'] + counts['ambig'] + counts['oth']), 4)
     counts['conflict'] = round(counts['ref'] / float(counts['alt'] + counts['ref'] + counts['ambig'] + counts['oth']), 4)
-    call = call1 and call2
+    if not call1 or not call2:
+        call = False
+    elif call1 == call2:
+        call = call1
+    elif call1 == "default":
+        call = call2
+    else:
+        call = call1
+
     note = note1
     if note != "" and note2 != "":
         note += ";" + note2
@@ -989,10 +1010,12 @@ def classify_constellations(in_fasta, list_constellation_files, constellation_na
             best_support = 0
             best_conflict = 1
             best_counts = None
+            best_call = False
             scores = {}
             children = {}
             for constellation in constellation_dict:
                 constellation_name = name_dict[constellation]
+                logging.debug("Consider constellation %s" %constellation_name)
                 parents = []
                 if not constellation_name:
                     continue
@@ -1015,20 +1038,25 @@ def classify_constellations(in_fasta, list_constellation_files, constellation_na
                     children[parent].append(constellation)
 
                 if call:
+                    logging.debug("Have call for %s" %constellation_name)
                     if call_all:
+                        if call != "default":
+                            constellation_name = "%s %s" %(call, constellation_name)
                         lineages.append(constellation_name)
                         names.append(constellation)
                     elif constellation in children and best_constellation in children[constellation]:
-                        continue
+                        logging.debug("Ignore as parent of best constellation")
                     elif (not best_constellation) \
                         or (counts['support'] > best_support) \
                         or (counts['support'] == best_support and counts['conflict'] < best_conflict)\
                         or (counts['support'] == best_support and counts['conflict'] == best_conflict and counts['rules'] > best_counts["rules"])\
                         or (best_constellation in parents):
                         best_constellation = constellation
+                        logging.debug("Set best constellation %s" %best_constellation)
                         best_support = counts['support']
                         best_conflict = counts['conflict']
                         best_counts = counts
+                        best_call = call
 
                 if interspersion:
                     if counts["alt"] > 1:
@@ -1042,7 +1070,11 @@ def classify_constellations(in_fasta, list_constellation_files, constellation_na
                                                        counts['oth'], counts['rules'], counts['support'],
                                                           counts['conflict'], call, constellation, note))
             if not call_all and best_constellation:
-                lineages.append(name_dict[best_constellation])
+                if best_call != "default":
+                    best_constellation_name = "%s %s" % (best_call, name_dict[best_constellation])
+                else:
+                    best_constellation_name = name_dict[best_constellation]
+                lineages.append(best_constellation_name)
                 names.append(best_constellation)
 
             out_entries = [record.id, "|".join(lineages), "|".join([mrca_lineage_dict[n] for n in names])]

scorpio/tests/type_constellations_test.py

@@ -147,9 +147,9 @@ def test_parse_json_in():
     assert len([v for v in variant_list if v["type"] == "del"]) == 3
     assert len([v for v in variant_list if v["type"] == "aa"]) == 15
     assert name == "Lineage_X"
-    assert rules["min_alt"] == 4
-    assert rules["max_ref"] == 6
-    assert rules["s:E484K"] == "alt"
+    assert rules["default"]["min_alt"] == 4
+    assert rules["default"]["max_ref"] == 6
+    assert rules["default"]["s:E484K"] == "alt"
     assert mrca_lineage == "B.1.1.7"
     assert incompatible_lineages == "A|B.1.351"
 
@@ -162,7 +162,7 @@ def test_parse_csv_in():
     assert len([v for v in variant_list if v["type"] == "del"]) == 3
     assert len([v for v in variant_list if v["type"] == "aa"]) == 15
     assert name == "lineage_X"
-    assert rules["s:E484K"] == "alt"
+    assert rules["default"]["s:E484K"] == "alt"
 
 
 def test_parse_textfile_in():
@@ -178,8 +178,8 @@ def test_parse_textfile_in():
 def test_parse_variants_in():
     in_files = ["%s/lineage_X.json" % data_dir, "%s/lineage_X.csv" % data_dir, "%s/lineage_X.txt" % data_dir]
     expect_names = ["Lineage_X", "lineage_X", "lineage_X"]
-    rule_dict_json = {"min_alt": 4, "max_ref": 6, "s:E484K": "alt"}
-    rule_dict_csv = {"s:E484K": "alt"}
+    rule_dict_json = {"default": {"min_alt": 4, "max_ref": 6, "s:E484K": "alt"}}
+    rule_dict_csv = {"default": {"s:E484K": "alt"}}
     rule_dict_txt = None
     expect_rules = [rule_dict_json, rule_dict_csv, rule_dict_txt]
 
@@ -248,8 +248,8 @@ def test_count_and_classify():
     oth_string = "gaaattcgcccgta-gctcgcaatag"
     seqs = [Seq(ref_string), Seq(alt_string), Seq(alt_plus_string), Seq(oth_string)]
 
-    rules = {"min_alt": 1, "max_ref": 1, "snp2": "alt"}
-    expect_classify = [False, False, True, False]
+    rules = {"default": {"min_alt": 1, "max_ref": 1, "snp2": "alt"}}
+    expect_classify = [False, False, "default", False]
     expect_counts = [{"ref": 5, "alt": 0, "ambig": 0, "oth": 1, "rules": 0, 'substitution': {'ref': 4, 'alt': 0, 'ambig': 0, 'oth': 0}, 'indel': {'ref': 1, 'alt': 0, 'ambig': 0, 'oth': 1}, "support": 0.0, "conflict": 0.8333},
                      {"ref": 1, "alt": 4, "ambig": 0, "oth": 1, "rules": 0, 'substitution': {'ref': 1, 'alt': 3, 'ambig': 0, 'oth': 0}, 'indel': {'ref': 0, 'alt': 1, 'ambig': 0, 'oth': 1}, "support": 0.6667, "conflict": 0.1667},
                      {"ref": 0, "alt": 5, "ambig": 0, "oth": 1, "rules": 3, 'substitution': {'ref': 0, 'alt': 4, 'ambig': 0, 'oth': 0}, 'indel': {'ref': 0, 'alt': 1, 'ambig': 0, 'oth': 1}, "support": 0.8333, "conflict": 0.0},