def parse_gene_ref(ref_gene) :
#FIXME - maybe, if galaxy doesn't work out, figure out how to deal with multiple RefGene mapping formats?
fieldnames = ['geneName','name','chrom','strand','txStart','txEnd','cdsStart','cdsEnd','exonCount','exonStarts','exonEnds']
reader = DictReader(ref_gene,fieldnames=fieldnames,delimiter='\t')
gene_ref = dd(list)
for ref_dict in reader :
for k,v in ref_dict.items() :
# coerce numbers where possible
ref_dict[k] = parse_number(v)
# turn 'x,x,x,...' into a list
ref_dict['exonStarts'] = [parse_number(x) for x in ref_dict['exonStarts'].split(',')]
if ref_dict['exonStarts'][-1] == '' : ref_dict['exonStarts'].remove('')
ref_dict['exonEnds'] = [parse_number(x) for x in ref_dict['exonEnds'].split(',')]
if ref_dict['exonEnds'][-1] == '' : ref_dict['exonEnds'].remove('')
gene_ref[ref_dict['chrom']].append(ref_dict)
return gene_ref
extrasaction='ignore',
lineterminator='\n')
peaks_writer.writerow(dict([(k, k) for k in output_fields]))
unique_genes = set()
map_stats = dd(int)
for peak in peaks_reader:
# if this is a comment or header line get skip it
if peak[fieldnames[0]].startswith('#') or \
peak[fieldnames[0]] == fieldnames[0] or \
peak[fieldnames[0]].startswith('track') :
continue
# coerce values to numeric if possible
for k, v in peak.items():
peak[k] = parse_number(v)
# MACS output gives us summit
if opts.peaks_fmt == 'MACS':
peak_loc = peak[start_field] + peak['summit']
else: # peak assumed to be in the middle of the reported peak range
peak_loc = (peak[start_field] + peak[end_field]) / 2
chrom_genes = gene_ref[peak[chr_field]]
if len(chrom_genes) == 0:
sys.stderr.write(
'WARNING: peak chromosome %s not found in gene reference, skipping: %s\n'
% (peak[chr_field], peak))
continue
symbol_xref_map[rec['kgID']] = rec
output_fields = ['knownGeneID','geneSymbol']+fieldnames
peaks_writer = DictWriter(peak_output,output_fields,delimiter='\t',extrasaction='ignore',lineterminator='\n')
peaks_writer.writerow(dict([(k,k) for k in output_fields]))
unique_genes = set()
map_stats = dd(int)
for peak in peaks_reader :
# if this is a comment or header line get skip it
if peak[fieldnames[0]].startswith('#') or \
peak[fieldnames[0]] == fieldnames[0] or \
peak[fieldnames[0]].startswith('track') : continue
# coerce values to numeric if possible
for k,v in peak.items() : peak[k] = parse_number(v)
# MACS output gives us summit
if opts.peaks_fmt == 'MACS' :
peak_loc = peak[start_field]+peak['summit']
else : # peak assumed to be in the middle of the reported peak range
peak_loc = (peak[start_field]+peak[end_field])/2
chrom_genes = gene_ref[peak[chr_field]]
if len(chrom_genes) == 0 :
sys.stderr.write('WARNING: peak chromosome %s not found in gene reference, skipping: %s\n'%(peak[chr_field],peak))
continue
mapped = False
print '\nParsing %d rows from peak file and will provide update every %d rows'%(totalrows,interval)
for peak in peaks_reader :
rowcount+=1
if rowcount % interval ==0:
print 'Processing row %d out of %d...'%(rowcount,totalrows)
# if this is a comment or header line get skip it
#removed 'startswith' call so that this can work with tuples
if peak[fieldnames[0]][0]=='#' or \
peak[fieldnames[0]] == fieldnames[0] or \
peak[fieldnames[0]][0]=='track' : continue
# coerce values to numeric if possible
for k,v in peak.items() : peak[k] = parse_number(v)
# MACS output gives us summit
if opts.peaks_fmt == 'MACS' :
peak_loc = int(peak[start_field])+int(peak['summit'])
elif opts.peaks_fmt == 'GPS' :
#get position and also add in a real window
ch,mid,tot = peak['Position']##reader already parses this into tuple
peak['Position']=tot
peak_loc = int(mid)
peak[chr_field] = ch
peak[start_field] = peak_loc-125
peak[end_field] = peak_loc+125
else : # peak assumed to be in the middle of the reported peak range
peak_loc = (int(peak[start_field])+int(peak[end_field]))/2
def main():
opts, args = parser.parse_args(sys.argv[1:])
if len(args) < 3 :
parser.error('Must provide three filename arguments')
gene_ref = parse_gene_ref(args[0])
xref_fn = args[1]
peaks_fn = args[2]
if opts.peaks_fmt == 'MACS' :
peaks_reader_cls = MACSFile
chr_field, start_field, end_field = 'chr', 'start', 'end'
elif opts.peaks_fmt == 'BED' :
peaks_reader_cls = BEDFile
chr_field, start_field, end_field = 'chrom', 'chromStart', 'chromEnd'
else :
# should never happen
fieldnames = []
#peaks_reader = DictReader(open(args[1]),fieldnames=fieldnames,delimiter='\t')
peaks_reader = peaks_reader_cls(peaks_fn)
# default output format:
if opts.peak_output :
peak_output = open(opts.peak_output,'w')
else :
peak_output = sys.stdout
fieldnames = peaks_reader.FIELD_NAMES
if opts.detail :
fieldnames += ["peak loc","dist from feature","map type","map subtype"]#"score"
output_fields = ['knownGeneID']+fieldnames
# see if the user wants gene symbols too
# TODO - actually make this an option, or make it required
opts.symbol_xref = xref_fn
if opts.symbol_xref :
kgXref_fieldnames = ['kgID','mRNA','spID','spDisplayID','geneSymbol','refseq','protAcc','description']
symbol_xref_reader = DictReader(open(opts.symbol_xref),fieldnames=kgXref_fieldnames,delimiter='\t')
symbol_xref_map = {}
for rec in symbol_xref_reader :
symbol_xref_map[rec['kgID']] = rec
output_fields = ['knownGeneID','geneSymbol']+fieldnames
peaks_writer = DictWriter(peak_output,output_fields,delimiter='\t',extrasaction='ignore',lineterminator='\n')
peaks_writer.writerow(dict([(k,k) for k in output_fields]))
unique_genes = set()
map_stats = dd(int)
for peak in peaks_reader :
# if this is a comment or header line get skip it
if peak[fieldnames[0]].startswith('#') or \
peak[fieldnames[0]] == fieldnames[0] or \
peak[fieldnames[0]].startswith('track') : continue
# coerce values to numeric if possible
for k,v in peak.items() : peak[k] = parse_number(v)
# MACS output gives us summit
if opts.peaks_fmt == 'MACS' :
peak_loc = peak[start_field]+peak['summit']
else : # peak assumed to be in the middle of the reported peak range
peak_loc = (peak[start_field]+peak[end_field])/2
chrom_genes = gene_ref[peak[chr_field]]
if len(chrom_genes) == 0 :
sys.stdout.write('WARNING: peak chromosome %s not found in gene reference, skipping: %s\n'%(peak[chr_field],peak))
continue
mapped = False
# walk through the genes for this chromosome
for gene in chrom_genes :
# reusable dictionary for output
out_d = {}.fromkeys(output_fields,0)
out_d.update(peak)
out_d['map type'] = ''
out_d['chromo'] = peak[chr_field]
out_d['peak loc'] = peak_loc
# determine intervals for promoter, gene, and downstream
if gene['strand'] == '+' :
promoter_coords = max(gene['txStart']-1-opts.upst_win,0), gene['txStart']-1
if opts.tss :
gene_coords = gene['txStart'], min(gene['txEnd'],gene['txStart']+opts.dnst_win)
downstream_coords = gene['txEnd']+1,gene['txStart']+opts.dnst_win
else :
gene_coords = gene['txStart'], gene['txEnd']
downstream_coords = gene['txEnd']+1, gene['txEnd']+1+opts.dnst_win
else :
promoter_coords = gene['txEnd']+1, gene['txEnd']+1+opts.upst_win # +1 because we're using 1 based indexing
if opts.tss :
gene_coords = max(gene['txStart'],gene['txEnd']-opts.upst_win), gene['txEnd']
downstream_coords = gene['txEnd']-1-opts.dnst_win, gene['txStart']-1 # -1 because we're using 1 based indexing
else :
gene_coords = gene['txStart'], gene['txEnd']
downstream_coords = gene['txStart']-1-opts.dnst_win, gene['txStart']-1 # -1 because we're using 1 based indexing
# check for promoter
if peak_loc >= promoter_coords[0] and peak_loc <= promoter_coords[1] :
out_d['map type'] = 'promoter'
out_d['dist from feature'] = peak_loc - promoter_coords[1] if gene['strand'] == '+' else promoter_coords[0] - peak_loc
# check for gene
elif peak_loc >= gene_coords[0] and peak_loc <= gene_coords[1] :
# check for intron/exon
exon_coords = zip(gene['exonStarts'],gene['exonEnds'])
in_exon = False
for st,en in exon_coords :
if peak_loc >= st and peak_loc <= en :
in_exon = True
break
out_d['map type'] = 'gene'
out_d['map subtype'] = 'exon' if in_exon else 'intron'
#Commented out to keep score reported in bed file - AJD 7/29/14
# score = (peak-TSS)/(TSE-TSS) - peak distance from TSS as fraction of length of gene
#gene_len = float(gene_coords[1]-gene_coords[0])
#out_d['score'] = (peak_loc-gene_coords[0])/gene_len if gene['strand'] == '+' else (gene_coords[1]-peak_loc)/gene_len
# distance calculated from start of gene
out_d['dist from feature'] = peak_loc - promoter_coords[1] if gene['strand'] == '+' else promoter_coords[0] - peak_loc
map_stats[out_d['map subtype']] += 1
# check for downstream
elif peak_loc >= downstream_coords[0] and peak_loc <= downstream_coords[1] :
out_d['map type'] = 'after'
if opts.tss :
out_d['dist from feature'] = peak_loc - gene_coords[0] if gene['strand'] == '+' else gene_coords[1] - peak_loc
else :
out_d['dist from feature'] = peak_loc - downstream_coords[0] if gene['strand'] == '+' else downstream_coords[1] - peak_loc
# does not map to this gene
else :
pass
# map type is not blank if we mapped to something
if out_d['map type'] != '' :
#out_d = {'knownGeneID':gene['name']}
out_d['knownGeneID'] = gene['name']
if opts.symbol_xref :
out_d['geneSymbol'] = symbol_xref_map[gene['name']]['geneSymbol']
peaks_writer.writerow(out_d)
mapped = True
# reset map_type
out_d['map type'] = ''
if not mapped :
if opts.intergenic :
out_d['knownGeneID'] = 'None'
out_d['geneSymbol'] = 'None'
out_d['map type'] = 'intergenic'
peaks_writer.writerow(out_d)
map_stats['intergenic'] += 1
if peak_output != sys.stdout:
peak_output.close()
