def getBioseqFromSetList( self, lSets ):
header = "%s::%i %s " % ( lSets[0].name, lSets[0].id, lSets[0].seqname )
sequence = ""
lSortedSets = SetUtils.getSetListSortedByIncreasingMinThenMax( lSets )
if not lSets[0].isOnDirectStrand():
lSortedSets.reverse()
for iSet in lSortedSets:
header += "%i..%i," % ( iSet.getStart(), iSet.getEnd() )
sequence += self.getSubSequence( iSet.seqname, iSet.getStart(), iSet.getEnd() )
return Bioseq( header[:-1], sequence )
def getIdentityFromPathList( lPaths, checkSubjects=True ):
if len( PathUtils.getListOfDistinctQueryNames( lPaths ) ) > 1:
msg = "ERROR: try to compute identity from Paths with different queries"
sys.stderr.write( "%s\n" % msg ); sys.stderr.flush()
raise Exception
if checkSubjects and len( PathUtils.getListOfDistinctSubjectNames( lPaths ) ) > 1:
msg = "ERROR: try to compute identity from Paths with different subjects"
sys.stderr.write( "%s\n" % msg ); sys.stderr.flush()
raise Exception
identity = 0
lMergedPaths = PathUtils.mergePathsInListUsingQueryCoordsOnly( lPaths )
lQuerySets = PathUtils.getSetListFromQueries( lMergedPaths )
lMergedQuerySets = SetUtils.mergeSetsInList( lQuerySets )
totalLengthOnQry = SetUtils.getCumulLength( lMergedQuerySets )
for iPath in lMergedPaths:
identity += iPath.identity * iPath.getLengthOnQuery()
weightedIdentity = identity / float(totalLengthOnQry)
if weightedIdentity < 0 or weightedIdentity > 100:
msg = "ERROR: weighted identity '%.2f' outside range" % ( weightedIdentity )
sys.stderr.write( "%s\n" % msg ); sys.stderr.flush()
raise Exception
return weightedIdentity
lSetIdToRemovePaths = []
count = 0
# for each path ID
for id in lPathId:
string = "processing path '%i'..." % ( id )
if qtype == "path":
lPaths = qtablePathAdaptator.getPathListFromId( id )
lQuerySets = PathUtils.getSetListFromQueries( lPaths )
elif qtype == "set":
lQuerySets = qtableSetAdaptator.getSetListFromId( id )
lQuerySets.sort()
qmin, qmax = SetUtils.getListBoundaries( lQuerySets )
qmin = qmin - 1
qmax = qmax + 1
if stype == "path":
lPaths = stablePathAdaptator.getPathListOverlappingQueryCoord( lQuerySets[0].seqname.split()[0], qmin, qmax )
lSubjectSets = PathUtils.getSetListFromQueries( lPaths )
elif stype == "set":
lSubjectSets = stableSetAdaptator.getSetListFromQueryCoord( lQuerySets[0].seqname.split()[0], qmin, qmax )
if verbose > 1:
print "----------------------------------------"
if len(lSubjectSets) > 0:
if verbose > 1:
print "annot:"
def getLengthOnQueryFromPathList( lPaths ):
lSets = PathUtils.getSetListFromQueries( lPaths )
lMergedSets = SetUtils.mergeSetsInList( lSets )
length = SetUtils.getCumulLength( lMergedSets )
return length
