def remove_wrong_periodicity(input,crossdata):
"""
Remove PacbPs with wrong periodicity from crossdata
@type input: dict
@param input: input <dict data structure> with lists of Orfs
@type crossdata: dict
@param crossdata: crossdata <dict data structure>
"""
for (geneQ,geneS) in crossdata.keys():
keys = crossdata[(geneQ,geneS)]['accepted_pacbs'].keys()
for key in keys:
pacbp = crossdata[(geneQ,geneS)]['accepted_pacbs'][key]
if pacbp.__class__.__name__ == 'PacbPORF':
# pacbp is already a PacbPORF
pacbporf = pacbp
else:
(bitscore,length,orfQid,orfSid) = key
orfQ = input[geneQ]['orfs'].get_orf_by_id(orfQid)
orfS = input[geneS]['orfs'].get_orf_by_id(orfSid)
pacbporf = pacb.conversion.pacbp2pacbporf(pacbp,orfQ,orfS)
# get aligned DNA sequences from the pacbporf
(dnaQ,dnaS) = pacbporf.get_aligned_dna_sequences()
# do coding_sequence_periodicity_test
(_test1,_test2) = coding_sequence_periodicity_test(dnaQ,dnaS)
if (_test1,_test2) != (True, True):
# store to rejected
crossdata[(geneQ,geneS)]['rejected_pacbs_wp'][key] = pacbporf
# remove from accepted_pacbps dict
del( crossdata[(geneQ,geneS)]['accepted_pacbs'][key] )
# okay, return the updated crossdata
return crossdata
def is_coding(self):
""" Does this aligment likely represent a coding (DNA) alignment? """
Qseq,Sseq = self.get_unextended_aligned_dna_sequences()
####### ERRONEOUS! why??
if len(Qseq) != len(Sseq):
print "\n" , self
print len(Qseq), "[UNEQUAL LENGTH]" ,len(Sseq)
print Qseq[0:30], "...", Qseq[-30:]
print Sseq[0:30], "...", Sseq[-30:]
####### ERRONEOUS! why??
return coding_sequence_periodicity_test(Qseq,Sseq)
def is_coding(self):
""" Does this aligment likely represent a coding (DNA) alignment? """
Qseq, Sseq = self.get_unextended_aligned_dna_sequences()
####### ERRONEOUS! why??
if len(Qseq) != len(Sseq):
print "\n", self
print len(Qseq), "[UNEQUAL LENGTH]", len(Sseq)
print Qseq[0:30], "...", Qseq[-30:]
print Sseq[0:30], "...", Sseq[-30:]
####### ERRONEOUS! why??
return coding_sequence_periodicity_test(Qseq, Sseq)
