def assemble_seq(readid2seq, junc_seq, tmp_file_path):
match = 2
mismatch = -1
scoring = swalign.NucleotideScoringMatrix(match, mismatch)
sw = swalign.LocalAlignment(
scoring) # you can also choose gap penalties, etc...
hout = open(tmp_file_path + ".tmp3.assemble_input.fa", 'w')
for tid in sorted(readid2seq):
print >> hout, '>' + tid
print >> hout, readid2seq[tid]
hout.close()
hout = open(tmp_file_path + ".tmp3.assemble_output.fq", 'w')
sret = subprocess.call(
["fml-asm", tmp_file_path + ".tmp3.assemble_input.fa"], stdout=hout)
hout.close()
if sret != 0:
print >> sys.stderr, "fml-asm error, error code: " + str(sret)
sys.exit()
line_num = 0
temp_contig = ""
with open(tmp_file_path + ".tmp3.assemble_output.fq", 'r') as hin:
for line in hin:
line_num = line_num + 1
if line_num % 4 == 2:
tseq = line.rstrip('\n')
aln_1 = sw.align(tseq, junc_seq)
if aln_1.score >= 35:
ttcontig = tseq[aln_1.r_end:]
if len(ttcontig) > len(temp_contig): temp_contig = ttcontig
aln_2 = sw.align(tseq, my_seq.reverse_complement(junc_seq))
if aln_2.score >= 35:
ttcontig = my_seq.reverse_complement(tseq[:aln_2.r_pos])
if len(ttcontig) > len(temp_contig): temp_contig = ttcontig
# subprocess.call(["rm", "-rf", tmp_file_path + ".tmp3.assemble_input.fa"])
# subprocess.call(["rm", "-rf", tmp_file_path + ".tmp3.assemble_output.fq"])
return temp_contig
def generate_contig(input_file, output_file, tumor_bp_file, tumor_bam,
reference_genome, min_contig_length):
tumor_bp_db = pysam.TabixFile(tumor_bp_file)
readid2key = {}
with open(input_file, 'r') as hin:
for line in hin:
F = line.rstrip('\n').split('\t')
if F[0] == "Chr": continue
tabixErrorFlag = 0
try:
records = tumor_bp_db.fetch(F[0], int(F[1]) - 1, int(F[1]) + 1)
except Exception as inst:
print >> sys.stderr, "%s: %s" % (type(inst), inst.args)
tabixErrorMsg = str(inst.args)
tabixErrorFlag = 1
if tabixErrorFlag == 0:
for record_line in records:
record = record_line.split('\t')
if record[0] == F[0] and (int(record[1]) + 1) == int(
F[1]) and record[3] == F[2] and record[4] == F[3]:
for readid in record[5].split(';'):
readid2key[re.sub(r'/\d$', '',
readid)] = ','.join(F[:4])
bamfile = pysam.Samfile(tumor_bam, "rb")
hout = open(output_file + ".tmp2.contig.unsorted", 'w')
for read in bamfile.fetch():
if read.qname in readid2key:
flags = format(int(read.flag), "#014b")[:1:-1]
# skip supplementary alignment
if flags[8] == "1" or flags[11] == "1": continue
# skip duplicated reads
if flags[10] == "1": continue
print >> hout, readid2key[read.qname] + '\t' + read.qname + (
"/1" if flags[6] == "1" else "/2") + '\t' + read.query_sequence
hout.close()
hout = open(output_file + ".tmp2.contig.sorted", 'w')
subprocess.call(["sort", "-k1,1", output_file + ".tmp2.contig.unsorted"],
stdout=hout)
hout.close()
temp_key = ""
temp_id2seq = {}
temp_junc_seq = ""
key2contig = {}
with open(output_file + ".tmp2.contig.sorted") as hin:
for line in hin:
F = line.rstrip('\n').split('\t')
if temp_key != F[0]:
if len(temp_id2seq) > 0:
key2contig[temp_key] = assemble_seq(
temp_id2seq, temp_junc_seq, output_file)
temp_key = F[0]
temp_id2seq = {}
FF = temp_key.split(',')
if FF[2] == "+":
temp_junc_seq = my_seq.get_seq(reference_genome, FF[0],
int(FF[1]) - 20, int(FF[1]))
else:
temp_junc_seq = my_seq.reverse_complement(
my_seq.get_seq(reference_genome, FF[0], int(FF[1]),
int(FF[1]) + 20))
temp_id2seq[F[1]] = F[2]
if len(temp_id2seq) > 0:
key2contig[temp_key] = assemble_seq(temp_id2seq, temp_junc_seq,
output_file)
hout = open(output_file, 'w')
with open(input_file, 'r') as hin:
for line in hin:
F = line.rstrip('\n').split('\t')
key = ','.join(F[:4])
if key not in key2contig: continue
contig = key2contig[key]
if len(contig) < min_contig_length: continue
# if contig[:8] != F[3][:8]: continue
print >> hout, '\t'.join(F) + '\t' + contig
hout.close()
def parse_bp_from_bam(input_bam, output_file, key_seq_size,
min_major_clip_size, max_minor_clip_size):
bamfile = pysam.Samfile(input_bam, "rb")
hout = open(output_file, "w")
# maybe add the regional extraction of bam files
for read in bamfile.fetch():
# get the flag information
flags = format(int(read.flag), "#014b")[:1:-1]
# skip if not aligned
if flags[2] == "1": continue
# skip supplementary alignment
if flags[8] == "1" or flags[11] == "1": continue
# skip duplicated reads
if flags[10] == "1": continue
# no clipping
if len(read.cigar) == 1: continue
# get the clipping size in the both side
left_clipping = (read.cigar[0][1] if read.cigar[0][0] in [4, 5] else 0)
right_clipping = (read.cigar[len(read.cigar) - 1][1] if
read.cigar[len(read.cigar) - 1][0] in [4, 5] else 0)
if left_clipping < min_major_clip_size and right_clipping < min_major_clip_size:
continue
# get the alignment basic information
chr_current = bamfile.getrname(read.tid)
pos_current = int(read.pos + 1)
dir_current = ("-" if flags[4] == "1" else "+")
# when the right side is clipped...
if right_clipping >= min_major_clip_size:
clipLen_current = right_clipping
alignmentSize_current = read.alen
readLength_current = read.rlen
juncChr_current = chr_current
juncPos_current = pos_current + alignmentSize_current - 1
juncDir_current = "+"
juncseq_start = readLength_current - clipLen_current
juncseq_end = readLength_current - clipLen_current + key_seq_size
juncseq = read.seq[juncseq_start:juncseq_end]
print >> hout, '\t'.join([
juncChr_current,
str(juncPos_current - 1),
str(juncPos_current), juncDir_current, juncseq,
read.qname + ("/1" if flags[6] == "1" else "/2"),
str(read.mapq),
str(right_clipping),
str(alignmentSize_current)
])
if left_clipping >= min_major_clip_size:
clipLen_current = left_clipping
alignmentSize_current = read.alen
readLength_current = read.rlen
juncChr_current = chr_current
juncPos_current = pos_current
juncDir_current = "-"
juncseq_end = clipLen_current
juncseq_start = clipLen_current - key_seq_size
juncseq = my_seq.reverse_complement(
read.seq[juncseq_start:juncseq_end])
print >> hout, '\t'.join([
juncChr_current,
str(juncPos_current - 1),
str(juncPos_current), juncDir_current, juncseq,
read.qname + ("/1" if flags[6] == "1" else "/2"),
str(read.mapq),
str(left_clipping),
str(alignmentSize_current)
])
bamfile.close()
hout.close()
def filter_by_base_quality(input_file, output_file, tumor_bam, min_support_num,
permissible_range):
hout = open(output_file, 'w')
with open(input_file, 'r') as hin:
for line in hin:
F = line.rstrip('\n').split('\t')
tumor_bamfile = pysam.Samfile(tumor_bam, "rb")
key_baseq = []
tumor_var_read = 0
tabixErrorFlag = 0
try:
records = tumor_bamfile.fetch(F[0], max(int(F[1]) - 1, 0),
int(F[1]) + 1)
except Exception as inst:
print >> sys.stderr, "%s: %s" % (type(inst), inst.args)
tabixErrorMsg = str(inst.args)
tabixErrorFlag = 1
if tabixErrorFlag == 0:
for read in records:
flags = format(int(read.flag), "#014b")[:1:-1]
# skip if not aligned
if flags[2] == "1": continue
# skip supplementary alignment
if flags[8] == "1" or flags[11] == "1": continue
# skip duplicated reads
if flags[10] == "1": continue
# no clipping
if len(read.cigar) == 1: continue
left_clipping = (read.cigar[0][1]
if read.cigar[0][0] in [4, 5] else 0)
right_clipping = (read.cigar[len(read.cigar) - 1][1]
if read.cigar[len(read.cigar) -
1][0] in [4, 5] else 0)
#print right_clipping
if F[2] == "+":
if right_clipping < 2: continue
juncPos_current = int(read.pos + 1) + read.alen - 1
if int(juncPos_current) != int(F[1]): continue
juncseq_start = read.rlen - right_clipping
juncseq_end = min(juncseq_start + 8, read.rlen)
juncseq = read.seq[juncseq_start:juncseq_end]
#if numpy.mean(read.query_qualities[juncseq_start:juncseq_end]) < 10:
# continue
if juncseq in F[3]:
tumor_var_read += 1
key_baseq.append(
str(
numpy.mean(read.query_qualities[
juncseq_start:juncseq_end])))
if F[2] == "-":
if left_clipping < 2: continue
juncPos_current = int(read.pos + 1)
if int(juncPos_current) != int(F[1]): continue
juncseq_start = left_clipping
juncseq_end = max(left_clipping - 8, 0)
juncseq = my_seq.reverse_complement(
read.seq[juncseq_start:juncseq_end])
#if numpy.mean(read.query_qualities[juncseq_start:juncseq_end]) < 10:
# continue
if juncseq in F[3]:
tumor_var_read += 1
key_baseq.append(
str(
numpy.mean(read.query_qualities[
juncseq_end:juncseq_start])))
if tumor_var_read < min_support_num: continue
if numpy.median(list(map(float, key_baseq))) < 20: continue
if numpy.sort(list(map(float, key_baseq)))[-2] < 30: continue
print >> hout, F[0] + '\t' + F[1] + '\t' + F[2] + '\t' + F[
3] + '\t' + str(tumor_var_read)
hin.close()
hout.close()
def parse_bp_from_bam(input_bam, output_file, key_seq_size,
min_major_clip_size, max_minor_clip_size):
"""
function for getting breakpoints from BAM file
input:bam file
output:output_file + ".bp.tmp.txt"
breakpoint info (juncChr, juncPos-1, juncPos, Dir (right clipping:+, left clipping:-), Juncseq, ID + ("/1" or "/2"), MAPQ, Clipping size, AlignmentSize, Base quality of juncseq)
"""
bamfile = pysam.Samfile(input_bam, "rb")
hout = open(output_file, "w")
# maybe add the regional extraction of bam files
for read in bamfile.fetch():
# get the flag information
flags = format(int(read.flag), "#014b")[:1:-1]
# skip if not aligned
if flags[2] == "1": continue
# skip supplementary alignment
if flags[8] == "1" or flags[11] == "1": continue
# skip duplicated reads
if flags[10] == "1": continue
# no clipping
if len(read.cigar) == 1: continue
# get the clipping size in the both side
left_clipping = (read.cigar[0][1] if read.cigar[0][0] in [4, 5] else 0)
right_clipping = (read.cigar[len(read.cigar) - 1][1] if
read.cigar[len(read.cigar) - 1][0] in [4, 5] else 0)
if left_clipping < min_major_clip_size and right_clipping < min_major_clip_size:
continue
# get the alignment basic information
chr_current = bamfile.getrname(read.tid)
pos_current = int(read.pos + 1)
dir_current = ("-" if flags[4] == "1" else "+")
# when the right side is clipped...
if right_clipping >= min_major_clip_size:
clipLen_current = right_clipping
alignmentSize_current = read.alen
readLength_current = read.rlen
juncChr_current = chr_current
juncPos_current = pos_current + alignmentSize_current - 1
juncDir_current = "+"
juncseq_start = readLength_current - clipLen_current
juncseq_end = readLength_current - clipLen_current + key_seq_size
juncseq = read.seq[juncseq_start:juncseq_end]
#filter if base qualities of junction seq is low
#if numpy.mean(read.query_qualities[juncseq_start:juncseq_end]) < 10:
# continue
print >> hout, '\t'.join([
juncChr_current,
str(juncPos_current - 1),
str(juncPos_current), juncDir_current, juncseq,
read.qname + ("/1" if flags[6] == "1" else "/2"),
str(read.mapq),
str(right_clipping),
str(alignmentSize_current)
])
if left_clipping >= min_major_clip_size:
clipLen_current = left_clipping
alignmentSize_current = read.alen
readLength_current = read.rlen
juncChr_current = chr_current
juncPos_current = pos_current
juncDir_current = "-"
juncseq_end = clipLen_current
juncseq_start = clipLen_current - key_seq_size
juncseq = my_seq.reverse_complement(
read.seq[juncseq_start:juncseq_end])
#filter if base qualities of soft clipping part is low
#if numpy.mean(read.query_qualities[juncseq_start:juncseq_end])<10:
# continue
print >> hout, '\t'.join([
juncChr_current,
str(juncPos_current - 1),
str(juncPos_current), juncDir_current, juncseq,
read.qname + ("/1" if flags[6] == "1" else "/2"),
str(read.mapq),
str(left_clipping),
str(alignmentSize_current)
])
bamfile.close()
hout.close()