Add additional species that are now supported because of MHCflurry

griffithlab · Jul 29, 2020 · b648217 · b648217
1 parent d0f5222
commit b648217
Show file tree

Hide file tree

Showing 2 changed files with 74 additions and 19 deletions.
diff --git a/lib/calculate_reference_proteome_similarity.py b/lib/calculate_reference_proteome_similarity.py
@@ -18,12 +18,38 @@ def __init__(self, input_file, input_fasta, output_file, match_length=8, species
         self.file_type = file_type
         self.species_to_organism = {
             'human': 'Homo sapiens',
-            'mouse': 'Mus musculus',
+            'atlantic salmon': 'Salmo salar',
+            'black-headed spider monkey': 'Ateles fusciceps',
+            'blue monkey': 'Cercopithecus mitis',
+            'bonobo': 'Pan paniscus',
+            'bornean orangutan': 'Pongo pygmaeus',
+            'brown-mantled tamarin': 'Saguinus fuscicollis',
             'chimpanzee': 'Pan troglodytes',
-            'macaque': 'Macaca',
+            'common marmoset': 'Callithrix jacchus',
+            'common squirrel monkey': 'Saimiri sciureus',
+            'cottontop tamarin': 'Saguinus oedipus',
             'cow': 'Bos taurus',
+            'crab-eating macaque': 'Macaca fascicularis',
+            'dog': 'Canis lupus familiaris',
+            "Geoffroy's tamarin": 'Saguinus geoffroyi',
+            'golden lion tamarin': 'Leontopithecus rosalia',
+            'gorilla': 'Gorilla gorilla',
+            'grivet': 'Chlorocebus aethiops',
+            'hamadryas baboon': 'Papio hamadryas',
             'horse': 'Equus caballus',
+            'lar gibbon': 'Hylobates lar',
+            'mouse': 'Mus musculus',
+            'moustached tamarin': 'Saguinus mystax',
+            'olive baboon': 'Papio anubis',
             'pig': 'Sus scrofa',
+            'rainbow trout': 'Oncorhynchus mykiss',
+            'rhesus macaque': 'Macaca mulatta',
+            'sheep': 'Ovis aries',
+            'southern pig-tailed macaque': 'Macaca nemestrina',
+            'stump-tailed macaque': 'Macaca arctoides',
+            'white-faced saki': 'Pithecia pithecia',
+            'white-fronted spider monkey': 'Ateles belzebuth',
+            'yellow baboon': 'Papio cynocephalus',
         }
 
     def reference_match_headers(self):

diff --git a/lib/prediction_class.py b/lib/prediction_class.py
@@ -190,24 +190,53 @@ def check_alleles_valid(cls, alleles):
 
     @classmethod
     def species_for_allele(self, allele):
-        if allele.startswith('HLA'):
-            return 'human'
-        elif allele.startswith('Patr'):
-            return 'chimpanzee'
-        elif allele.startswith('Mamu'):
-            return 'macaque'
-        elif allele.startswith('BoLA'):
-            return 'cow'
-        elif allele.startswith('Eqca'):
-            return 'horse'
-        elif allele.startswith('H-2'):
-            return 'mouse'
-        elif allele.startswith('H2'):
-            return 'mouse'
-        elif allele.startswith('SLA'):
-            return 'pig'
+        allele_to_species = {
+            'HLA' : 'human',
+            'DP'  : 'human',
+            'DQ'  : 'human',
+            'DR'  : 'human',
+            'Atbe': 'white-fronted spider monkey',
+            'Atfu': 'black-headed spider monkey',
+            'BoLA': 'cow',
+            'Caja': 'common marmoset',
+            'Cemi': 'blue monkey',
+            'Chae': 'grivet',
+            'DLA' : 'dog',
+            'Eqca': 'horse',
+            'Gogo': 'gorilla',
+            'H-2' : 'mouse',
+            'H2'  : 'mouse',
+            'Hyla': 'lar gibbon',
+            'Lero': 'golden lion tamarin',
+            'Maar': 'stump-tailed macaque',
+            'Mafa': 'crab-eating macaque',
+            'Mamu': 'rhesus macaque',
+            'Mane': 'southern pig-tailed macaque',
+            'Onmy': 'rainbow trout',
+            'Ovar': 'sheep',
+            'Paan': 'olive baboon',
+            'Pacy': 'yellow baboon',
+            'Paha': 'hamadryas baboon',
+            'Papa': 'bonobo',
+            'Patr': 'chimpanzee',
+            'Pipi': 'white-faced saki',
+            'Popy': 'bornean orangutan',
+            'Safu': 'brown-mantled tamarin',
+            'Sage': "Geoffroy's tamarin",
+            'Samy': 'moustached tamarin',
+            'Saoe': 'cottontop tamarin',
+            'Sasa': 'atlantic salmon',
+            'Sasc': 'common squirrel monkey',
+            'SLA' : 'pig',
+        }
+
+        species = [v for k,v in allele_to_species.items() if allele.startswith(k)]
+        if len(species) == 1:
+            return species[0]
+        elif len(species) == 0:
+            raise Exception("Unable to determine species for allele {}".format(allele))
         else:
-            return 'human'
+            raise Exception("Multiple matching species found for allele {}".format(allele))
 
     @classmethod
     def parse_allele_cutoff_file(cls):