def integrator( self, variant, out_splts ) :
# get the variant field information from the first out_splt
# remember there may be multiple lines all with the same eqkey() evaluation,
# corresponding to different isoforms
for c in self.cols :
variant.fields[c] = self.nullify( out_splts[0][indexOf[c]] )
#make variant.isoforms out of each splt
for out_splt in out_splts :
iso = Isoform()
pp = out_splt[indexOf["proteinPosition"]]
if not pp == '' and not pp == 'NA':
splt = pp.split('/')
pos,tot = int(splt[0]), int(splt[1])
iso.fields["codon_pos"] = pos
iso.fields["codon_total"] = tot
iso.fields["accession"] = self.nullify(out_splt[indexOf["accession"]])
iso.fields["functionGVS"] = self.nullify(out_splt[indexOf["functionGVS"]])
iso.fields["polyPhen"] = self.nullify(out_splt[indexOf["polyPhen"]])
aas = out_splt[indexOf["aminoAcids"]].split(',')
if len(aas) == 2 :
iso.fields["ref_aa"] = aas[0]
iso.fields["mut_aa"] = aas[1]
variant.isoforms.append(iso)
iso.fields["gene"] = self.nullify( out_splt[indexOf["geneList"]] )
return variant
def integrator( self, variant, out_splts ) :
#get the variant field information from the first out_splt
#remember there may be multiple corresponding to different isoforms
for (c,dbc) in zip(self.cols,self.db_cols) :
variant.fields[c] = self.nullify( out_splts[0][indexOf[c]] )
#make variant.isoforms out of each splt
for out_splt in out_splts :
pp = out_splt[indexOf["proteinPosition"]]
pos,tot = -1,-1
if not pp == '' and not pp == 'NA':
splt = pp.split('/')
pos,tot = int(splt[0]), int(splt[1])
iso = Isoform()
iso.fields["accession"] = self.nullify(out_splt[indexOf["accession"]])
iso.fields["ss_functionGVS"] = self.nullify(out_splt[indexOf["functionGVS"]])
iso.fields["ss_polyPhen"] = self.nullify(out_splt[indexOf["polyPhen"]])
iso.fields["codon_pos"] = pos
iso.fields["codon_total"] = tot
aas = out_splt[indexOf["aminoAcids"]].split(',')
if len(aas) == 2 :
iso.fields["ref_aa"] = aas[0]
iso.fields["mut_aa"] = aas[1]
variant.isoforms.append(iso)
return variant
def varListIntegrator(self,variant,out_splt) :
novel_variant = "ref_aa" not in variant.fields
messages = []
if novel_variant :
aas = out_splt[ self.indexOf["old_aa/new_aa"] ].split('/')
if len(aas) == 2 :
variant.fields["ref_aa"] = aas[0]
variant.fields["mut_aa"] = aas[1]
iso_keys = ["transcript_id", \
"exon_rank", "effect"]
iso = Isoform()
for ik in iso_keys :
iso.fields[ik] = out_splt[ self.indexOf[ik] ]
codon_pos = out_splt[ self.indexOf["codon_num(cds)"] ]
if codon_pos :
iso.fields["codon_pos"] = int(codon_pos)
#else : iso.fields["codon_pos"] = codon_pos
cdssize = out_splt[ self.indexOf["cds_size"] ]
dbname = "codon_total"
if not cdssize == '' :
iso.fields[dbname] = int(cdssize)/3
#else : iso.fields[dbname] = cdssize
k = "gene_id"
ucsc_id = out_splt[self.indexOf[k]]
if not ucsc_id == '' :
query = "select id from Genes where ucsc_id = '%s'" \
% ucsc_id
gene_ids = self.conn.query(query)
if len(gene_ids) > 0 :
for gene_id in [int(row[0]) for row in gene_ids] :
clone = iso.clone()
clone.fields[k] = gene_id
variant.isoforms.append(clone)
else :
gid = importer.makeEmptyGene( self.conn, 'ucsc_id', ucsc_id )
iso.fields[k] = gid
messages.append( 'No gene id for ucsc_id=|%s|, create an empty gene, id: %d' % (ucsc_id,gid) )
variant.isoforms.append(iso)
else :
iso.fields[k] = -1
variant.isoforms.append(iso)
#print "varListIntegrator, variant got: %d isoforms" % len(variant.isoforms)
return '| \t |'.join(messages)