Merge pull request #24 from CapraLab/NeanderthalFixes

Misc tweaks to pdbmap library

lib/PDBMapProtein.py

@@ -344,7 +344,7 @@ def load_idmapping(cls, idmapping_fname):
                     hgnc = dbid  # for clarity
                     PDBMapProtein._unp2hgnc[unp] = hgnc
                     PDBMapProtein._hgnc2unp[hgnc] = unp
-                elif db == "Gene_ID":
+                elif db == "GeneID":
                     gene_id = dbid  # for clarity
                     PDBMapProtein._unp2gene_id[unp] = gene_id
                 elif db == "Ensembl_TRS":

lib/PDBMapTranscriptBase.py

@@ -109,6 +109,20 @@ def describe_transcript_differences(t1,t2):
         return differences            
 
 
+    @staticmethod
+    def analyze_transcript_differences(t1,t2):
+        """Return # of single point variants, and percent of mismatches, between 2 PDBMapTranscripts"""
+        if len(t1.aa_seq) != len(t2.aa_seq):
+            # If transcripts are not the same length, return the longer length, and 100% mismatch
+            return max(len(t1.aa_seq),len(t2.aa_seq)),1.0
+
+        differences = 0
+        for tpos in range(len(t1.aa_seq)):
+            if t1.aa_seq[tpos] != t2.aa_seq[tpos]:
+                differences += 1
+        return differences,float(differences)/float(len(t1.aa_seq))            
+
+
 # Main check
 if __name__== "__main__":
   LOGGER.critical("Class definition. Should not be called from command line.")