Flexible matching for constellation assignments; added additional sit…

…e for H3 motif; classify GISAID-exclusive strains

.gitignore

@@ -14,3 +14,4 @@ src/.orientdb_history
 build/
 dist/
 octofludb.egg-info/
+readme.txt

CHANGELOG.md

@@ -1,3 +1,10 @@
+v1.0.2 [2023-11-03]
+===================
+
+ * Make constellation determination more robust with partial, case-insensitive matching
+ * Correct H3 motif by adding missing site (159)
+ * Fix fetch-unclassified-swine.rq to include GISAID strains
+
 v1.0.1 [2022-02-16]
 ===================
 

octofludb/data/fetch-unclassified-swine.rq

@@ -1,10 +1,10 @@
 PREFIX f: <https://flu-crew.org/term/>
 
-SELECT DISTINCT ?gb ?seq
+SELECT DISTINCT ?seqid ?seq
 WHERE {
     ?sid f:host "swine" .
     ?sid f:has_segment ?gid .
-    ?gid f:genbank_id ?gb .
+    ?gid f:seqid ?seqid .
     ?gid f:dnaseq ?seq .
     MINUS {
         ?gid f:clade ?clade .

octofludb/formatting.py

@@ -72,7 +72,7 @@ def _make_constellations(rows: List[Tuple[str, str, str]]) -> List[Tuple[str, st
     segment_lookup = dict(PB2=0, PB1=1, PA=2, NP=3, M=4, MP=4, NS=5)
 
     clade_lookup = dict(
-        pdm="P", LAIV="V", TX98="V", TRIG="T", humanSeasonal="H", classicalSwine="C", avian="A"
+        pdm="P", LAIV="V", TRIG="T", humanSeasonal="H", classicalSwine="C", avian="A"
     )
 
     const: Dict[str, List[str]] = dict()
@@ -91,6 +91,11 @@ def _make_constellations(rows: List[Tuple[str, str, str]]) -> List[Tuple[str, st
 
         if clade in clade_lookup:
             char = clade_lookup[clade]
+        elif "-like" in clade.lower():
+            log(
+                f"{bad('WARNING:')} internal gene clade includes '-like' label ({clade}), assigning constellation character 'X'"
+            )
+            char = "X"
         else:
             # add flexible matching
             for c_lookup, c_letter in clade_lookup.items():
@@ -102,7 +107,7 @@ def _make_constellations(rows: List[Tuple[str, str, str]]) -> List[Tuple[str, st
                     break
             else:
                 log(
-                    f"{bad('WARNING:')} expected internal gene clade to be one of  'pdm', 'LAIV', 'TX98', 'TRIG', 'classicalSwine', 'humanSeasonal', or 'avian'. Found clade {clade}, assigning constellation character 'X'"
+                    f"{bad('WARNING:')} expected internal gene clade to be one of  'pdm', 'LAIV', 'TRIG', 'classicalSwine', 'humanSeasonal', or 'avian'. Found clade {clade}, assigning constellation character 'X'"
                 )
                 char = "X"
 

octofludb/ui.py

@@ -315,7 +315,7 @@ def upload_motifs(url: str, repo: str) -> List[str]:
     # find h3 motifs
     h3_motif_table = script.findMotifs(
         os.path.join(os.path.dirname(__file__), "data", "get-h3-swine.rq"),
-        ["h3_motif=145,155,156,158,189"],
+        ["h3_motif=145,155,156,158,159,189"],
         "H3",
         url=url,
         repo_name=repo,

octofludb/version.py

@@ -1 +1 @@
-__version__ = "1.0.1"
+__version__ = "1.0.2"