def match(self, query, scope, verbose=False):
out = []
for taxId, ncbiMatch in self._db.execute('select taxid, spname from species where spname like "%s%%";' % scope).fetchall(): # TODO where rank in (111,222,333)
match = sw.align( ncbiMatch, query)
out.append( (taxId, ncbiMatch, match, match.score) )
out.sort(key = lambda x:-(x[3]) )
if verbose:
for x in out[:3]:
print("--"*20)
print("%d %s" % (x[0], x[1]))
print( "score: %d matches: %d" % (x[2].score, x[2].matches))
x[2].dump()
else:
#out[0][2].dump()
pass
if not out:
return None
if out[0][2].matches < len(query)*self._requiredQueryIdentity:
return None
if len(out)>1 and not ( ( out[0][3] > out[1][3]) and (out[0][2].matches > out[1][2].matches ) ):
return None
return (out[0][0], out[0][1], float(out[0][2].matches)/len(query))
def findBestMatches(n1, n2):
bestMatches = {}
for a in n1:
best = None
best_b = None
for b in n2:
match = sw.align( removeUninformativeTerms(a), removeUninformativeTerms(b))
if best is None:
best = match
best_b = b
elif match.score > best.score:
best = match
best_b = b
bestMatches[a] = best_b
return bestMatches
def findMatchingName(name):
bestScore = 0.0
bestMatch = ""
bestMatchTaxId = 0
isBestMatchATie = False
for target, targetTaxId in speciesMapping.items():
match = sw.align(name, target)
#if match.score>0:
# print("{}\t{}".format( match.score, target) )
if match.score == bestScore: # tie
isBestMatchATie = True
elif match.score > bestScore and match.score > 0:
bestMatch = target
bestMatchTaxId = targetTaxId
bestScore = match.score
isBestMatchATie = False
return (bestMatch, bestMatchTaxId, isBestMatchATie)
def matchInexactNames(n1, n2):
n1 = set(n1)
n2 = set(n2)
bestMatches12 = findBestMatches(n1, n2)
bestMatches21 = findBestMatches(n2, n1)
matches = []
for a in n1:
b = bestMatches12[a]
if a == bestMatches21[b]:
#print("%s\t%s" % (a, b))
matches.append((a,b))
else:
print("**"*20)
print("%s -> %s" % (a, bestMatches12[a]))
sw.align(a, bestMatches12[a]).dump()
print("%s -> %s" % (b, bestMatches21[b]))
sw.align(b, bestMatches21[b]).dump()
print("--")
print("%s -> %s" % (bestMatches12[a], bestMatches21[bestMatches12[a]] ))
sw.align(bestMatches12[a], bestMatches21[bestMatches12[a]]).dump()
print("%s -> %s" % (bestMatches21[b], bestMatches12[bestMatches21[b]] ))
sw.align(bestMatches21[b], bestMatches12[bestMatches21[b]]).dump()
print("**"*20)
#revMatch = sw.align( best_b, a)
#print("--"*20)
#print(a)
#print(best_b)
#best.dump()
#revMatch.dump()
return matches