def get_gap_or_overlap_from_psl_file(psl_filename):
res = {}
with open(psl_filename) as infile:
while True:
line1 = infile.readline().strip()
if not line1: return res
line2 = infile.readline().strip()
if not line2: return res
e1 = psl_basics.read_psl_entry(line1)
e2 = psl_basics.read_psl_entry(line2)
prog = re.compile('\|(F\d+)[+-]')
m = prog.search(e1['qName'])
if not m:
print "problem reading query name from psl file"
sys.exit()
e1name = m.group(1)
m = prog.search(e2['qName'])
if not m:
print "problem reading query name from psl file"
sys.exit()
e2name = m.group(1)
if e1name != e2name:
print "problem: consecutive lines should have the same fusion ID"
sys.exit()
v = {}
v['base_overlap'] = psl_basics.query_coordinates_base_overlap_size(e1,e2)
v['region_overlap'] = psl_basics.query_coordinates_region_overlap_size(e1,e2)
v['gap'] = psl_basics.query_coordinates_gap_size(e1,e2)
res[e1name] = v
return res
def get_gap_or_overlap_from_psl_file(psl_filename):
res = {}
with open(psl_filename) as infile:
while True:
line1 = infile.readline().strip()
if not line1: return res
line2 = infile.readline().strip()
if not line2: return res
e1 = psl_basics.read_psl_entry(line1)
e2 = psl_basics.read_psl_entry(line2)
prog = re.compile('\|(F\d+)[+-]')
m = prog.search(e1['qName'])
if not m:
print "problem reading query name from psl file"
sys.exit()
e1name = m.group(1)
m = prog.search(e2['qName'])
if not m:
print "problem reading query name from psl file"
sys.exit()
e2name = m.group(1)
if e1name != e2name:
print "problem: consecutive lines should have the same fusion ID"
sys.exit()
v = {}
v['base_overlap'] = psl_basics.query_coordinates_base_overlap_size(e1,e2)
v['region_overlap'] = psl_basics.query_coordinates_region_overlap_size(e1,e2)
v['gap'] = psl_basics.query_coordinates_gap_size(e1,e2)
res[e1name] = v
return res
def main():
if len(sys.argv) < 2:
print sys.argv[0] + " <psl filename> <smoothing parameter (default 10)>"
sys.exit()
smoothing = 10
if len(sys.argv) == 3:
smoothing = int(sys.argv[2])
psl_filename = sys.argv[1]
with open(psl_filename) as fh:
for line in fh:
line = line.rstrip()
psl_entry = psl_basics.read_psl_entry(line)
gpd_line = psl_basics.convert_entry_to_genepred_line(psl_entry)
gpd_entry = genepred_basics.genepred_line_to_dictionary(gpd_line)
smoothed_gpd_entry = genepred_basics.smooth_gaps(gpd_entry,smoothing)
#get longest exon only
longest_exon = 0
best_start = 0
best_end = 0
for i in range(0,smoothed_gpd_entry['exonCount']):
exon_length = smoothed_gpd_entry['exonEnds'][i]-smoothed_gpd_entry['exonStarts'][i]
if exon_length > longest_exon:
longest_exon = exon_length
best_end = smoothed_gpd_entry['exonEnds'][i]
best_start = smoothed_gpd_entry['exonStarts'][i]
print psl_entry['tName'] + "\t" + str(best_start) + "\t" + str(best_end) + "\t" + psl_entry['qName'] + "\t" + str(psl_entry['qSize']) + "\t" + str(psl_entry['tSize']) + "\t" + str(longest_exon)
def make_best_continuous_alignment_bed(alignmentfname,bestalignmentbedfname):
bestcont = {}
of = open(bestalignmentbedfname,'w')
with open(alignmentfname) as f:
for line in f:
if re.match('^#',line): continue
pe = psl_basics.read_psl_entry(line)
gl = psl_basics.convert_entry_to_genepred_line(pe)
ge = genepred_basics.genepred_line_to_dictionary(gl)
ges = genepred_basics.smooth_gaps(ge,10)
for i in range(0,len(ges['exonStarts'])):
exonlen = ges['exonEnds'][i]-ges['exonStarts'][i]
if ges['name'] not in bestcont:
bestcont[ges['name']] = {}
if ges['chrom'] not in bestcont[ges['name']]:
entry = {}
entry['bestlen'] = 0
entry['start'] = 0
entry['end'] = 0
entry['target_length'] = 0
entry['query_length'] = 0
entry['matches'] = 0
bestcont[ges['name']][ges['chrom']] = entry
if exonlen > bestcont[ges['name']][ges['chrom']]['bestlen']:
bestcont[ges['name']][ges['chrom']]['bestlen'] = exonlen
bestcont[ges['name']][ges['chrom']]['start'] = ges['exonStarts'][i]
bestcont[ges['name']][ges['chrom']]['end'] = ges['exonEnds'][i]
bestcont[ges['name']][ges['chrom']]['target_length'] = pe['tSize']
bestcont[ges['name']][ges['chrom']]['query_length'] = pe['qSize']
bestcont[ges['name']][ges['chrom']]['matches'] = pe['matches']
for read in bestcont:
for tx in bestcont[read]:
of.write(tx + "\t" + str(bestcont[read][tx]['start']) + "\t" + str(bestcont[read][tx]['end']) + "\t" + read + "\t" + str(bestcont[read][tx]['query_length']) + "\t" + str(bestcont[read][tx]['target_length']) + "\t" + str(bestcont[read][tx]['bestlen']) + "\t" + str(bestcont[read][tx]['matches']) + "\n")
return
def main():
if len(sys.argv) < 2:
print sys.argv[0] + ' <psl filename> <smoothing size (optional)>'
sys.exit()
pslfilename = sys.argv[1]
smooth_size = 0
if len(sys.argv) == 3:
smooth_size = int(sys.argv[2])
with open(pslfilename) as infile:
for line in infile:
psl_entry = psl_basics.read_psl_entry(line)
genepred_line = psl_basics.convert_entry_to_genepred_line(psl_entry)
genepred_entry = genepred_basics.genepred_line_to_dictionary(genepred_line)
if smooth_size > 0:
genepred_entry = genepred_basics.smooth_gaps(genepred_entry,smooth_size)
for i in range(0,len(genepred_entry['exonStarts'])):
print genepred_entry['chrom'] + "\t" + str(genepred_entry['exonStarts'][i]+1) + "\t" + str(genepred_entry['exonEnds'][i]) + "\t" + genepred_entry['gene_name']
def make_fusion_genepred(psl_filename, min_gap_in_block_size, outfile):
i = 0
ofile = open(outfile, 'w')
fusion_gpd = {}
with open(psl_filename) as infile:
for line in infile:
m = i % 2
i += 1
entry = psl_basics.read_psl_entry(line)
gline = psl_basics.convert_entry_to_genepred_line(entry)
e = genepred_basics.genepred_line_to_dictionary(gline)
# m == 0 is left, m == 1 is right
look = re.search('\|(F[\d]+)([+-])', e['gene_name'])
fid = look.group(1)
fsidesign = look.group(2) #which side is the fusion on
e['cdsStart'] = e['txStart']
e['cdsEnd'] = e['txEnd']
f = genepred_basics.smooth_gaps(e, min_gap_in_block_size)
# get new transcript length
newlen = 0
for j in range(0, f['exonCount']):
newlen += f['exonEnds'][j] - f['exonStarts'][j]
# make a new name based on the new fusion site and new length
look = re.match('^([\d\.]+_)\d+(\|F\d+[+-].*)$', f['gene_name'])
v1 = look.group(1)
v2 = str(newlen)
v3 = look.group(2)
f['gene_name'] = v1 + v2 + v3
gpline = genepred_basics.genepred_entry_to_genepred_line(f)
ofile.write(gpline + "\n")
if fid not in fusion_gpd:
n = [0, 0]
fusion_gpd[fid] = n
fusion_gpd[fid][m] = [
f, fsidesign
] #send back the genepred and the side the fusion is on
ofile.close()
return fusion_gpd
def make_fusion_genepred(psl_filename,min_gap_in_block_size,outfile):
i = 0
ofile = open(outfile,'w')
fusion_gpd = {}
with open(psl_filename) as infile:
for line in infile:
m = i % 2
i += 1
entry = psl_basics.read_psl_entry(line)
gline = psl_basics.convert_entry_to_genepred_line(entry)
e = genepred_basics.genepred_line_to_dictionary(gline)
# m == 0 is left, m == 1 is right
look = re.search('\|(F[\d]+)([+-])',e['gene_name'])
fid = look.group(1)
fsidesign = look.group(2) #which side is the fusion on
e['cdsStart'] = e['txStart']
e['cdsEnd'] = e['txEnd']
f = genepred_basics.smooth_gaps(e,min_gap_in_block_size)
# get new transcript length
newlen = 0
for j in range(0,f['exonCount']): newlen += f['exonEnds'][j] - f['exonStarts'][j]
# make a new name based on the new fusion site and new length
look = re.match('^([\d\.]+_)\d+(\|F\d+[+-].*)$',f['gene_name'])
v1 = look.group(1)
v2 = str(newlen)
v3 = look.group(2)
f['gene_name'] = v1 + v2 + v3
gpline = genepred_basics.genepred_entry_to_genepred_line(f)
ofile.write(gpline+"\n")
if fid not in fusion_gpd:
n = [0,0]
fusion_gpd[fid] = n
fusion_gpd[fid][m] = [f, fsidesign] #send back the genepred and the side the fusion is on
ofile.close()
return fusion_gpd
def main():
if len(sys.argv) < 2:
print sys.argv[0] + ' <psl filename> <smoothing size (optional)>'
sys.exit()
pslfilename = sys.argv[1]
smooth_size = 0
if len(sys.argv) == 3:
smooth_size = int(sys.argv[2])
with open(pslfilename) as infile:
for line in infile:
psl_entry = psl_basics.read_psl_entry(line)
genepred_line = psl_basics.convert_entry_to_genepred_line(
psl_entry)
genepred_entry = genepred_basics.genepred_line_to_dictionary(
genepred_line)
if smooth_size > 0:
genepred_entry = genepred_basics.smooth_gaps(
genepred_entry, smooth_size)
for i in range(0, len(genepred_entry['exonStarts'])):
print genepred_entry['chrom'] + "\t" + str(
genepred_entry['exonStarts'][i] + 1) + "\t" + str(
genepred_entry['exonEnds']
[i]) + "\t" + genepred_entry['gene_name']
def make_fusion_genepred(psl_filename,fid_coords,min_gap_in_block_size,outfile):
i = 0
ofile = open(outfile,'w')
fusion_gpd = {}
with open(psl_filename) as infile:
for line in infile:
m = i % 2
i += 1
entry = psl_basics.read_psl_entry(line)
gline = psl_basics.convert_entry_to_genepred_line(entry)
d = genepred_basics.genepred_line_to_dictionary(gline)
# m == 0 is left, m == 1 is right
look = re.search('\|(F[\d]+)([+-])',d['gene_name'])
fid = look.group(1)
fsidesign = look.group(2) #which side is the fusion on
if fid not in fid_coords: continue # we only are working on the ones that met criteria for short reads
#fix our end points based on short read defined fusions
side = 'left'
if m == 1: side = 'right'
[chr, coord, dir] = fid_coords[fid][side]
coord = int(coord)
# fix our endpoints
e = {} # our corrected genepred
newpos = 0 #this will be used to update the
if fsidesign == '-':
#print 'right ' +str(coord) + " " + str(d['exonEnds'][d['exonCount']-1])
# if the new txEnd is greater than the last exon, we need to move it out
if coord == d['exonEnds'][d['exonCount']-1]:
e = d
#print "right side stays the same"
elif coord > d['exonEnds'][d['exonCount']-1]:
e = genepred_basics.right_extend_genepred(d,coord)
#print "right extend transcript"
#everything is all set
else:
e = genepred_basics.right_trim_genepred(d,coord)
#print "right shorten transcript"
newpos = e['txEnd']
else:
# end point is on the left side of the alignment
#print 'left ' + str(coord) + " " + str(d['exonStarts'][0])
if coord-1 == d['exonStarts'][0]:
e = d
#print "left stays the same"
if coord-1 < d['exonStarts'][0]:
e = genepred_basics.left_extend_genepred(d,coord-1)
#print "left extend transcript"
else:
e = genepred_basics.left_trim_genepred(d,coord-1)
#print "left trim transcript"
newpos = e['txStart']
# we treat cdsStart and cdsEnd the same as tx
e['cdsStart'] = e['txStart']
e['cdsEnd'] = e['txEnd']
f = genepred_basics.smooth_gaps(e,min_gap_in_block_size)
# get new transcript length
newlen = 0
for j in range(0,f['exonCount']): newlen += f['exonEnds'][j] - f['exonStarts'][j]
# make a new name based on the new fusion site and new length
look = re.match('^([\d\.]+_)\d+(\|F\d+[+-])\d+(\|.*)$',f['gene_name'])
v1 = look.group(1)
v2 = str(newlen)
v3 = look.group(2)
v4 = str(newpos)
v5 = look.group(3)
f['gene_name'] = v1 + v2 + v3 + v4 + v5
gpline = genepred_basics.genepred_entry_to_genepred_line(f)
ofile.write(gpline+"\n")
if fid not in fusion_gpd:
n = [0,0]
fusion_gpd[fid] = n
fusion_gpd[fid][m] = [f, fsidesign] #send back the genepred and the side the fusion is on
ofile.close()
return fusion_gpd
def process_temp_list(temp_list, gname_list, gene_start_pos_list,
gene_end_pos_list, gene_max_end_pos_list):
ref_stat = 0
# Select best ref_ls from single alignemt
for result_ls in temp_list:
stat = float(result_ls[0]) / float(result_ls[10])
if stat > ref_stat:
ref_stat = stat
ref_str = '\t'.join(result_ls)
fusion_flag = False
# Select best ref_ls from two alignemts
# qstart/end index: 11/12
# tstart/end index: 15/16
# tname index: 13
temp_list_len = len(temp_list)
for idx in range(temp_list_len):
# Check numbe rof matches
gnames_1 = get_gnames(temp_list[idx][13], int(temp_list[idx][15]),
int(temp_list[idx][16]), gname_list,
gene_start_pos_list, gene_end_pos_list,
gene_max_end_pos_list)
if (int(temp_list[idx][0]) < fusion_segment_len_threshold):
continue
line1 = "\t".join(temp_list[idx])
entry1 = psl_basics.read_psl_entry(line1)
for idx_2 in range(idx + 1, temp_list_len):
# Check number of matches
if (int(temp_list[idx_2][0]) < fusion_segment_len_threshold):
continue
if (temp_list[idx][13] == temp_list[idx_2][13]):
#print ">" + str(temp_list)
gnames_2 = get_gnames(temp_list[idx_2][13],
int(temp_list[idx_2][15]),
int(temp_list[idx_2][16]), gname_list,
gene_start_pos_list, gene_end_pos_list,
gene_max_end_pos_list)
# Check if they share same gene locus
if ((len(gnames_1) > 0) and (len(gnames_2) > 0)):
if (len(gnames_1 & gnames_2) != 0):
continue
else:
tstart = max(int(temp_list[idx][15]),
int(temp_list[idx_2][15]))
tend = min(int(temp_list[idx][16]),
int(temp_list[idx_2][16]))
# Note: it is expected intron shorter than L_junction_limit are not splited by blat
if ((tstart - tend) < L_junction_limit):
continue
# Note: query coordinates have special handling in neg strands
# Check psl format on genome USCSC website
line2 = "\t".join(temp_list[idx_2])
entry2 = psl_basics.read_psl_entry(line2)
baseoverlap = psl_basics.query_coordinates_base_overlap_size(
entry1, entry2)
basegap = psl_basics.query_coordinates_gap_size(entry1, entry2)
if (baseoverlap):
if (baseoverlap <= fusion_overlap_threshold):
qlen = sum(entry1['blockSizes']) - baseoverlap
qlen_2 = sum(entry2['blockSizes']) - baseoverlap
if ((qlen < fusion_segment_len_threshold)
or (qlen_2 < fusion_segment_len_threshold)):
continue
else:
continue
elif (basegap > fusion_gap_threshold):
continue
stat = (float(temp_list[idx][0]) +
float(temp_list[idx_2][0])) / float(temp_list[idx][10])
if stat > ref_stat:
fusion_psl.write('\t'.join(temp_list[idx]) + '\n')
fusion_psl.write('\t'.join(temp_list[idx_2]) + '\n')
fusion_flag = True
if (fusion_flag):
fusion_psl_single.write(ref_str + '\n')
return ""
else:
return (ref_str)
def main():
if len(sys.argv) != 5:
print sys.argv[
0] + ' <IDP temp dir> <junction (i.e. chr1:1000-/chr2:1000+)> <range> <output dir>'
sys.exit()
inputdir = sys.argv[1].rstrip('/')
psl_filename = inputdir + '/' + 'LR.psl_fusion_pair_filtered'
map_filename = inputdir + '/uniqueness/fusion_read.map'
fasta_filename = inputdir + '/uniqueness/unmapped_shortread.fa'
txn_filename = inputdir + '/uniqueness/txn.map'
junction_abbreviation = sys.argv[2]
myrange = int(sys.argv[3])
outdir = sys.argv[4].rstrip('/')
if not os.path.isdir(outdir):
print "make directory " + outdir
os.mkdir(outdir)
of1 = open(outdir + "/fusion_coordiante.txt", 'w')
of1.write(junction_abbreviation + "\t" + str(myrange) + "\n")
of1.close()
m = re.match('([^:]+):(\d+)([+-])\/([^:]+):(\d+)([+-])',
junction_abbreviation)
chr1 = m.group(1)
coo1 = int(m.group(2))
dir1 = m.group(3)
chr2 = m.group(4)
coo2 = int(m.group(5))
dir2 = m.group(6)
of = open(outdir + "/long_read_query_locations.txt", 'w')
oleft = open(outdir + "/long_read_left.bed", 'w')
oright = open(outdir + "/long_read_right.bed", 'w')
oleftgpd = open(outdir + "/long_read_left.gpd", 'w')
orightgpd = open(outdir + "/long_read_right.gpd", 'w')
# Work through psl file for long reads
lrcnt = 0
with open(psl_filename) as f:
while True:
l1 = f.readline().rstrip()
if not l1:
break
l2 = f.readline().rstrip()
if not l2:
break
e1 = psl_basics.read_psl_entry(l1)
e2 = psl_basics.read_psl_entry(l2)
g1 = genepred_basics.smooth_gaps(
genepred_basics.genepred_line_to_dictionary(
psl_basics.convert_entry_to_genepred_line(e1)), 30)
g2 = genepred_basics.smooth_gaps(
genepred_basics.genepred_line_to_dictionary(
psl_basics.convert_entry_to_genepred_line(e2)), 30)
if check_coordiantes(e1, e2, chr1, coo1, dir1, chr2, coo2, dir2,
myrange):
oleftgpd.write(
genepred_basics.genepred_entry_to_genepred_line(g1) + "\n")
orightgpd.write(
genepred_basics.genepred_entry_to_genepred_line(g2) + "\n")
lrcnt += 1
of.write(e1["qName"] + "\t" + str(e1['qStart']+1) + "\t" + \
str(e1['qEnd']) + "\t" + dir1 + "\t" + \
str(e2['qStart']+1) + "\t" + str(e2['qEnd']) + "\t" + dir2 + "\n")
for i in range(0, g1['exonCount']):
if g1["exonEnds"][i] - g1["exonStarts"][i] >= 30:
oleft.write(g1["chrom"] + "\t" + str(g1["exonStarts"][i]+1) + "\t" + \
str(g1["exonEnds"][i]) + "\t" + e1["qName"] + "\n")
for i in range(0, g2['exonCount']):
if g2["exonEnds"][i] - g2["exonStarts"][i] >= 30:
oright.write(g2["chrom"] + "\t" + str(g2["exonStarts"][i]+1) + "\t" + \
str(g2["exonEnds"][i]) + "\t" + e1["qName"] + "\n")
elif check_coordiantes(e1, e2, chr2, coo2, opposite(dir2), chr1,
coo1, opposite(dir1), myrange):
oleftgpd.write(
genepred_basics.genepred_entry_to_genepred_line(g2) + "\n")
orightgpd.write(
genepred_basics.genepred_entry_to_genepred_line(g1) + "\n")
lrcnt += 1
of.write(e1["qName"] + "\t" + str(e1['qStart']+1) + "\t" + \
str(e1['qEnd']) + "\t" + opposite(dir2) + "\t" + \
str(e2['qStart']+1) + "\t" + str(e2['qEnd']) + "\t" + opposite(dir1) + "\n")
for i in range(0, g1['exonCount']):
if g1["exonEnds"][i] - g1["exonStarts"][i] >= 30:
oright.write(g1["chrom"] + "\t" + str(g1["exonStarts"][i]+1) + "\t" + \
str(g1["exonEnds"][i]) + "\t" + e1["qName"] + "\n")
for i in range(0, g2['exonCount']):
if g2["exonEnds"][i] - g2["exonStarts"][i] >= 30:
oleft.write(g2["chrom"] + "\t" + str(g2["exonStarts"][i]+1) + "\t" + \
str(g2["exonEnds"][i]) + "\t" + e1["qName"] + "\n")
of.close()
oleft.close()
oright.close()
oleftgpd.close()
orightgpd.close()
print str(lrcnt) + " long reads found supporting the fusion"
#Work through fusion read map for short reads
rnames = {}
seenhit = {}
with open(map_filename) as inf:
for line in inf:
f = line.rstrip().split("\t")
srname = f[0]
loc = f[2]
m = re.search('^([^:]+):.*\/([^:]+):', loc)
srchr1 = m.group(1)
srchr2 = m.group(2)
m = re.search(
'[,:](-?\d+)-(-?\d+)([+-])\/[^:]+:(-?\d+)-(-?\d+),?.*([+-])',
loc)
srcoo1start = int(m.group(1))
srcoo1finish = int(m.group(2))
srdir1 = m.group(3)
srcoo2start = int(m.group(4))
srcoo2finish = int(m.group(5))
srdir2 = m.group(6)
m = re.search
srcooleft = srcoo1finish
if srdir1 == '-':
srcooleft = srcoo1start
srcooright = srcoo2start
if srdir2 == '-':
srcooright = srcoo2finish
#print srchr1 + "\t" + srchr2 + "\t" + str(srcooleft) + "\t" + str(srcooright)
if srdir1 == dir1 and srchr1 == chr1 and srdir2 == dir2 and srchr2 == chr2 and srcooleft == coo1 and srcooright == coo2:
rnames[srname] = {}
rnames[srname]['left'] = srchr1 + "\t" + str(
srcoo1start) + "\t" + str(srcoo1finish) + "\t" + srname
rnames[srname]['right'] = srchr2 + "\t" + str(
srcoo2start) + "\t" + str(srcoo2finish) + "\t" + srname
if srname not in seenhit:
seenhit[srname] = 0
seenhit[srname] += 1
if srdir1 == opposite(
dir1
) and srchr1 == chr2 and srdir2 == opposite(
dir2
) and srchr2 == chr1 and srcooleft == coo2 and srcooright == coo1:
rnames[srname] = {}
rnames[srname]['left'] = srchr2 + "\t" + str(
srcoo2start) + "\t" + str(srcoo2finish) + "\t" + srname
rnames[srname]['right'] = srchr1 + "\t" + str(
srcoo1start) + "\t" + str(srcoo1finish) + "\t" + srname
if srname not in seenhit:
seenhit[srname] = 0
seenhit[srname] += 1
print "found " + str(len(rnames)) + " short reads"
for srname in seenhit:
if seenhit[srname] > 1:
print "removing " + srname
del rnames[srname]
print "found " + str(
len(rnames)) + " short reads with no multihits among fusions"
validreads = {}
with open(fasta_filename) as inf:
for line in inf:
m = re.match('^>(.*)$', line.rstrip())
if m:
validreads[m.group(1)] = 1
namelist = rnames.keys()
for rname in namelist:
if rname not in validreads:
print "removing " + rname
del rnames[rname]
print "found " + str(
len(rnames)) + " unique short reads with no hits in the genome"
with open(txn_filename) as inf:
for line in inf:
f = line.rstrip().split("\t")
if f[0] in rnames:
print "removing " + f[0]
del rnames[f[0]]
print "found " + str(
len(rnames)
) + " unique short reads with no hits in the genome or transcriptome"
oleft = open(outdir + "/short_read_left.bed", 'w')
oright = open(outdir + "/short_read_right.bed", 'w')
for rname in rnames:
oleft.write(rnames[rname]['left'] + "\n")
oright.write(rnames[rname]['right'] + "\n")
oleft.close()
oright.close()
def process_temp_list(temp_list,
gname_list, gene_start_pos_list,
gene_end_pos_list, gene_max_end_pos_list):
ref_stat = 0
# Select best ref_ls from single alignemt
for result_ls in temp_list:
stat = float(result_ls[0])/float(result_ls[10])
if stat > ref_stat:
ref_stat = stat
ref_str = '\t'.join(result_ls)
fusion_flag = False
# Select best ref_ls from two alignemts
# qstart/end index: 11/12
# tstart/end index: 15/16
# tname index: 13
temp_list_len = len(temp_list)
for idx in range(temp_list_len):
# Check numbe rof matches
gnames_1 = get_gnames(temp_list[idx][13], int(temp_list[idx][15]), int(temp_list[idx][16]),
gname_list, gene_start_pos_list, gene_end_pos_list, gene_max_end_pos_list)
if (int(temp_list[idx][0]) < fusion_segment_len_threshold):
continue
line1 = "\t".join(temp_list[idx])
entry1 = psl_basics.read_psl_entry(line1)
for idx_2 in range(idx + 1, temp_list_len):
# Check number of matches
if (int(temp_list[idx_2][0]) < fusion_segment_len_threshold):
continue
if (temp_list[idx][13] == temp_list[idx_2][13]):
#print ">" + str(temp_list)
gnames_2 = get_gnames(temp_list[idx_2][13], int(temp_list[idx_2][15]), int(temp_list[idx_2][16]),
gname_list, gene_start_pos_list, gene_end_pos_list, gene_max_end_pos_list)
# Check if they share same gene locus
if ((len(gnames_1) > 0) and (len(gnames_2) > 0)):
if (len(gnames_1 & gnames_2) != 0):
continue
else:
tstart = max(int(temp_list[idx][15]), int(temp_list[idx_2][15]))
tend = min(int(temp_list[idx][16]), int(temp_list[idx_2][16]))
# Note: it is expected intron shorter than L_junction_limit are not splited by blat
if ((tstart - tend) < L_junction_limit):
continue
# Note: query coordinates have special handling in neg strands
# Check psl format on genome USCSC website
line2 = "\t".join(temp_list[idx_2])
entry2 = psl_basics.read_psl_entry(line2)
baseoverlap = psl_basics.query_coordinates_base_overlap_size(entry1,entry2)
basegap = psl_basics.query_coordinates_gap_size(entry1,entry2)
if (baseoverlap):
if (baseoverlap <= fusion_overlap_threshold):
qlen = sum(entry1['blockSizes']) - baseoverlap
qlen_2 = sum(entry2['blockSizes']) - baseoverlap
if ((qlen < fusion_segment_len_threshold) or
(qlen_2 < fusion_segment_len_threshold)):
continue
else:
continue
elif (basegap > fusion_gap_threshold):
continue
stat = (float(temp_list[idx][0]) + float(temp_list[idx_2][0]))/float(temp_list[idx][10])
if stat > ref_stat:
fusion_psl.write('\t'.join(temp_list[idx]) + '\n')
fusion_psl.write('\t'.join(temp_list[idx_2]) + '\n')
fusion_flag = True
if (fusion_flag):
fusion_psl_single.write(ref_str + '\n')
return ""
else:
return (ref_str)
def main():
if len(sys.argv) != 5:
print sys.argv[0] + ' <IDP temp dir> <junction (i.e. chr1:1000-/chr2:1000+)> <range> <output dir>'
sys.exit()
inputdir = sys.argv[1].rstrip('/')
psl_filename = inputdir+'/'+'LR.psl_fusion_pair_filtered'
map_filename = inputdir+'/uniqueness/fusion_read.map'
fasta_filename = inputdir+'/uniqueness/unmapped_shortread.fa'
txn_filename = inputdir+'/uniqueness/txn.map'
junction_abbreviation = sys.argv[2]
myrange = int(sys.argv[3])
outdir = sys.argv[4].rstrip('/')
if not os.path.isdir(outdir):
print "make directory "+outdir
os.mkdir(outdir)
of1 = open(outdir+"/fusion_coordiante.txt",'w')
of1.write(junction_abbreviation+"\t"+str(myrange)+"\n")
of1.close()
m = re.match('([^:]+):(\d+)([+-])\/([^:]+):(\d+)([+-])',junction_abbreviation)
chr1 = m.group(1)
coo1 = int(m.group(2))
dir1 = m.group(3)
chr2 = m.group(4)
coo2 = int(m.group(5))
dir2 = m.group(6)
of = open(outdir+"/long_read_query_locations.txt",'w')
oleft = open(outdir+"/long_read_left.bed",'w')
oright = open(outdir+"/long_read_right.bed",'w')
oleftgpd = open(outdir+"/long_read_left.gpd",'w')
orightgpd = open(outdir+"/long_read_right.gpd",'w')
# Work through psl file for long reads
lrcnt = 0
with open(psl_filename) as f:
while True:
l1 = f.readline().rstrip()
if not l1:
break
l2 = f.readline().rstrip()
if not l2:
break
e1 = psl_basics.read_psl_entry(l1)
e2 = psl_basics.read_psl_entry(l2)
g1 = genepred_basics.smooth_gaps(genepred_basics.genepred_line_to_dictionary(psl_basics.convert_entry_to_genepred_line(e1)),30)
g2 = genepred_basics.smooth_gaps(genepred_basics.genepred_line_to_dictionary(psl_basics.convert_entry_to_genepred_line(e2)),30)
if check_coordiantes(e1,e2,chr1,coo1,dir1,chr2,coo2,dir2,myrange):
oleftgpd.write(genepred_basics.genepred_entry_to_genepred_line(g1)+"\n")
orightgpd.write(genepred_basics.genepred_entry_to_genepred_line(g2)+"\n")
lrcnt += 1
of.write(e1["qName"] + "\t" + str(e1['qStart']+1) + "\t" + \
str(e1['qEnd']) + "\t" + dir1 + "\t" + \
str(e2['qStart']+1) + "\t" + str(e2['qEnd']) + "\t" + dir2 + "\n")
for i in range(0,g1['exonCount']):
if g1["exonEnds"][i]-g1["exonStarts"][i] >= 30:
oleft.write(g1["chrom"] + "\t" + str(g1["exonStarts"][i]+1) + "\t" + \
str(g1["exonEnds"][i]) + "\t" + e1["qName"] + "\n")
for i in range(0,g2['exonCount']):
if g2["exonEnds"][i]-g2["exonStarts"][i] >= 30:
oright.write(g2["chrom"] + "\t" + str(g2["exonStarts"][i]+1) + "\t" + \
str(g2["exonEnds"][i]) + "\t" + e1["qName"] + "\n")
elif check_coordiantes(e1,e2,chr2,coo2,opposite(dir2),chr1,coo1,opposite(dir1),myrange):
oleftgpd.write(genepred_basics.genepred_entry_to_genepred_line(g2)+"\n")
orightgpd.write(genepred_basics.genepred_entry_to_genepred_line(g1)+"\n")
lrcnt += 1
of.write(e1["qName"] + "\t" + str(e1['qStart']+1) + "\t" + \
str(e1['qEnd']) + "\t" + opposite(dir2) + "\t" + \
str(e2['qStart']+1) + "\t" + str(e2['qEnd']) + "\t" + opposite(dir1) + "\n")
for i in range(0,g1['exonCount']):
if g1["exonEnds"][i]-g1["exonStarts"][i] >= 30:
oright.write(g1["chrom"] + "\t" + str(g1["exonStarts"][i]+1) + "\t" + \
str(g1["exonEnds"][i]) + "\t" + e1["qName"] + "\n")
for i in range(0,g2['exonCount']):
if g2["exonEnds"][i]-g2["exonStarts"][i] >= 30:
oleft.write(g2["chrom"] + "\t" + str(g2["exonStarts"][i]+1) + "\t" + \
str(g2["exonEnds"][i]) + "\t" + e1["qName"] + "\n")
of.close()
oleft.close()
oright.close()
oleftgpd.close()
orightgpd.close()
print str(lrcnt) + " long reads found supporting the fusion"
#Work through fusion read map for short reads
rnames = {}
seenhit = {}
with open(map_filename) as inf:
for line in inf:
f = line.rstrip().split("\t")
srname = f[0]
loc = f[2]
m = re.search('^([^:]+):.*\/([^:]+):',loc)
srchr1 = m.group(1)
srchr2 = m.group(2)
m = re.search('[,:](-?\d+)-(-?\d+)([+-])\/[^:]+:(-?\d+)-(-?\d+),?.*([+-])',loc)
srcoo1start = int(m.group(1))
srcoo1finish = int(m.group(2))
srdir1 = m.group(3)
srcoo2start = int(m.group(4))
srcoo2finish = int(m.group(5))
srdir2 = m.group(6)
m = re.search
srcooleft = srcoo1finish
if srdir1 == '-':
srcooleft = srcoo1start
srcooright = srcoo2start
if srdir2 == '-':
srcooright = srcoo2finish
#print srchr1 + "\t" + srchr2 + "\t" + str(srcooleft) + "\t" + str(srcooright)
if srdir1 == dir1 and srchr1 == chr1 and srdir2 == dir2 and srchr2 == chr2 and srcooleft == coo1 and srcooright == coo2:
rnames[srname] = {}
rnames[srname]['left'] = srchr1 + "\t" + str(srcoo1start) + "\t" + str(srcoo1finish) + "\t" + srname
rnames[srname]['right'] = srchr2 + "\t" + str(srcoo2start) + "\t" + str(srcoo2finish) + "\t" + srname
if srname not in seenhit:
seenhit[srname] = 0
seenhit[srname] += 1
if srdir1 == opposite(dir1) and srchr1 == chr2 and srdir2 == opposite(dir2) and srchr2 == chr1 and srcooleft == coo2 and srcooright == coo1:
rnames[srname] = {}
rnames[srname]['left'] = srchr2 + "\t" + str(srcoo2start) + "\t" + str(srcoo2finish) + "\t" + srname
rnames[srname]['right'] = srchr1 + "\t" + str(srcoo1start) + "\t" + str(srcoo1finish) + "\t" + srname
if srname not in seenhit:
seenhit[srname] = 0
seenhit[srname] += 1
print "found "+str(len(rnames))+" short reads"
for srname in seenhit:
if seenhit[srname] > 1:
print "removing " + srname
del rnames[srname]
print "found "+str(len(rnames))+" short reads with no multihits among fusions"
validreads = {}
with open(fasta_filename) as inf:
for line in inf:
m = re.match('^>(.*)$',line.rstrip())
if m:
validreads[m.group(1)] = 1
namelist = rnames.keys()
for rname in namelist:
if rname not in validreads:
print "removing " + rname
del rnames[rname]
print "found "+str(len(rnames))+" unique short reads with no hits in the genome"
with open(txn_filename) as inf:
for line in inf:
f = line.rstrip().split("\t")
if f[0] in rnames:
print "removing " + f[0]
del rnames[f[0]]
print "found "+str(len(rnames))+" unique short reads with no hits in the genome or transcriptome"
oleft = open(outdir+"/short_read_left.bed",'w')
oright = open(outdir+"/short_read_right.bed",'w')
for rname in rnames:
oleft.write(rnames[rname]['left']+"\n")
oright.write(rnames[rname]['right']+"\n")
oleft.close()
oright.close()
def make_fusion_genepred(psl_filename,fid_coords,min_gap_in_block_size,outfile):
i = 0
ofile = open(outfile,'w')
fusion_gpd = {}
with open(psl_filename) as infile:
for line in infile:
m = i % 2
i += 1
entry = psl_basics.read_psl_entry(line)
gline = psl_basics.convert_entry_to_genepred_line(entry)
d = genepred_basics.genepred_line_to_dictionary(gline)
# m == 0 is left, m == 1 is right
look = re.search('\|(F[\d]+)([+-])',d['gene_name'])
fid = look.group(1)
fsidesign = look.group(2) #which side is the fusion on
if fid not in fid_coords: continue # we only are working on the ones that met criteria for short reads
#fix our end points based on short read defined fusions
side = 'left'
if m == 1: side = 'right'
[chr, coord, dir] = fid_coords[fid][side]
coord = int(coord)
# fix our endpoints
e = {} # our corrected genepred
newpos = 0 #this will be used to update the
if fsidesign == '-':
#print 'right ' +str(coord) + " " + str(d['exonEnds'][d['exonCount']-1])
# if the new txEnd is greater than the last exon, we need to move it out
if coord == d['exonEnds'][d['exonCount']-1]:
e = d
#print "right side stays the same"
elif coord > d['exonEnds'][d['exonCount']-1]:
e = genepred_basics.right_extend_genepred(d,coord)
#print "right extend transcript"
#everything is all set
else:
e = genepred_basics.right_trim_genepred(d,coord)
#print "right shorten transcript"
newpos = e['txEnd']
else:
# end point is on the left side of the alignment
#print 'left ' + str(coord) + " " + str(d['exonStarts'][0])
if coord-1 == d['exonStarts'][0]:
e = d
#print "left stays the same"
if coord-1 < d['exonStarts'][0]:
e = genepred_basics.left_extend_genepred(d,coord-1)
#print "left extend transcript"
else:
e = genepred_basics.left_trim_genepred(d,coord-1)
#print "left trim transcript"
newpos = e['txStart']
# we treat cdsStart and cdsEnd the same as tx
e['cdsStart'] = e['txStart']
e['cdsEnd'] = e['txEnd']
f = genepred_basics.smooth_gaps(e,min_gap_in_block_size)
# get new transcript length
newlen = 0
for j in range(0,f['exonCount']): newlen += f['exonEnds'][j] - f['exonStarts'][j]
# make a new name based on the new fusion site and new length
look = re.match('^([\d\.]+_)\d+(\|F\d+[+-])\d+(\|.*)$',f['gene_name'])
v1 = look.group(1)
v2 = str(newlen)
v3 = look.group(2)
v4 = str(newpos)
v5 = look.group(3)
f['gene_name'] = v1 + v2 + v3 + v4 + v5
gpline = genepred_basics.genepred_entry_to_genepred_line(f)
ofile.write(gpline+"\n")
if fid not in fusion_gpd:
n = [0,0]
fusion_gpd[fid] = n
fusion_gpd[fid][m] = [f, fsidesign] #send back the genepred and the side the fusion is on
ofile.close()
return fusion_gpd
