Try to handle species in refseq database.

workflows/rnaseqdenovo/bin/blast2annot.py

@@ -62,7 +62,7 @@ def spFromHit( alignment ):
       @param alignment : [Bio::blast::Record::Alignment] the alignment.
     """
     species = None
-    if re.match(".*OS=.*", alignment.hit_def) is not None:
+    if re.match(".*OS=.*", alignment.hit_def) is not None: # refseq_prot, swissprot
         try:
             matches = re.search("OS=([^=]+)", alignment.hit_def)
             if matches is not None:
@@ -73,16 +73,30 @@ def spFromHit( alignment ):
                     species = clean_matches.group(1)
         except: 
             pass
-    else :                
-        try:
-            matches = re.search('\[([^\]]+)\]', alignment.hit_def)
-            species = matches.group(1)
-        except: 
-            species = None
-            pass
+    else :
+        #refseq_genomic >gi|76496352|ref|NT_165333.1|NT_165333 Mus musculus genomic sequence, ENCODE region ENm001
+        #refseq_rna >gi|31341369|ref|NM_176670.2| Bos taurus ATP synthase, H+ transporting, mitochondrial F1 complex, delta subunit (ATP5D), mRNA
+        #refseq_prot >gi|66818355|ref|XP_642837.1| hypothetical protein DDB_G0276911 [Dictyostelium discoideum AX4]
+        if re.match("gi\|\d+\|ref\|\S+\|", alignment.hit_id) is not None: #if refseq
+            try:
+                matches = re.search('\[([^\]]+)\]', alignment.hit_def) #si refseq_prot espece entre []
+                species = matches.group(1)
+            except: # refseq_genomic ou refseq_rna
+                matches = re.search("gi\|\d+\|ref\|\S+\|\s(\S+\s\S+)", alignment.hit_def)
+                if matches is not None:
+                    species = matches.group(1)
+                
+        else :
+            try:
+                matches = re.search('\[([^\]]+)\]', alignment.hit_def)
+                species = matches.group(1)
+            except: 
+                species = None
+                pass
     # Others databases are not implemented
     return species
 
+
 def geneNameFromHit( alignment, gene_name_from_accession ):
     """
      @summary : Returns the subject's name.