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
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:"
SetUtils.showList( lQuerySets )
lQuerySets = SetUtils.mergeSetsInList( lQuerySets )
qsize = SetUtils.getCumulLength( lQuerySets )
string += "\nannot size: %i" % ( qsize )
if verbose > 1:
print "annot size=", qsize
if verbose > 1:
print "simple repeats list:"
SetUtils.showList( lSubjectSets )
lSubjectSets = SetUtils.mergeSetsInList( lSubjectSets )
ssize = SetUtils.getCumulLength( lSubjectSets )
string += "\nsimple repeats size: %i" % ( ssize )
osize = SetUtils.getOverlapLengthBetweenLists( lQuerySets, lSubjectSets )
string += "\noverlap size: %i" % ( osize )
string += "\nremains %i" % ( qsize - osize )
if verbose > 1:
def getLengthOnQueryFromPathList( lPaths ):
lSets = PathUtils.getSetListFromQueries( lPaths )
lMergedSets = SetUtils.mergeSetsInList( lSets )
length = SetUtils.getCumulLength( lMergedSets )
return length
