def deal_translocation_unbalance(bam, ref, svinfo, is_single, readLength,
tempDir, insertSize, minmapq,
is_multmapfilter):
freq = svinfo.freq
chrom = svinfo.chrom
start = svinfo.start
end = svinfo.end
trans_chr, trans_start, trans_end = svinfo.trans_chrom, svinfo.trans_start, svinfo.trans_end
info = "%s:%s-%s" % (chrom, start, end)
svtype = "trans_unbalance"
if is_single:
reads_type6_left, reads_type6_right, reads_type7, filtered_read_num = pos_type_classify(
bam, chrom, start, end, is_single, readLength, tempDir)
trans_reads_type6_left, trans_reads_type6_right, trans_reads_type7, filtered_read_num = pos_type_classify(
bam,
trans_chr,
trans_start,
trans_end,
is_single,
readLength,
tempDir,
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
type_reads_num = len(reads_type6_left) + len(reads_type6_right) + len(
reads_type7)
freq = freq * (type_reads_num + filtered_read_num) / type_reads_num
if freq > 1:
freq = 1
modify_reads_type6 = deal_type6(ref,
trans_reads_type6_left,
trans_reads_type6_left,
freq,
trans_start,
trans_end,
svtype,
subPos=info)
add_reads_type7 = deal_type7(reads_type7, freq, svtype)
return modify_reads_type6, [], add_reads_type7
else:
reads_type1_left, reads_type1_right, reads_type2_left, reads_type2_right, reads_type3_left, reads_type3_right, reads_type4, reads_type5_left, reads_type5_right, filtered_read_num = pos_type_classify(
bam, chrom, start, end, is_single, readLength, tempDir)
trans_reads_type1_left, trans_reads_type1_right, trans_reads_type2_left, trans_reads_type2_right, trans_reads_type3_left, trans_reads_type3_right, trans_reads_type4, trans_reads_type5_left, trans_reads_type5_right, filtered_read_num = pos_type_classify(
bam,
trans_chr,
trans_start,
trans_end,
is_single,
readLength,
tempDir,
extension=insertSize[1],
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
# exit()
type_reads_num = len(reads_type1_left) + len(reads_type1_right) + len(
reads_type2_left) + len(reads_type2_right) + len(
reads_type3_left) + len(reads_type3_right) + len(
reads_type4) + len(reads_type5_left) + len(
reads_type5_right)
freq = freq * (type_reads_num * 2 +
filtered_read_num) / (type_reads_num * 2)
modify_reads_type1 = deal_type1(ref,
trans_reads_type1_left, [],
freq,
trans_start,
trans_end,
svtype,
subPos=info)
modify_reads_type2 = deal_type2(ref, [],
trans_reads_type2_left,
freq,
trans_start,
trans_end,
svtype,
subPos=info)
modify_reads_type3 = deal_type3(trans_reads_type3_left,
trans_reads_type3_left,
freq,
insertSize,
trans_start,
trans_end,
svtype,
supple1=reads_type3_left,
supple2=reads_type3_right,
subPos=info)
if modify_reads_type3 is False:
return False, "The reads are not satisfied to modify, see more details in log file"
add_reads_type4 = deal_type4(reads_type4,
freq,
svtype,
insertSize=insertSize)
modify_reads_type5 = deal_type5(ref,
trans_reads_type5_left,
trans_reads_type5_left,
freq,
trans_start,
trans_end,
svtype,
subPos=info)
return modify_reads_type1 + modify_reads_type2 + modify_reads_type3 + modify_reads_type5, [], add_reads_type4
def deal_translocation_chrom(bam, ref, svinfo, is_single, read_length, tempDir,
insertSize, minmapq, is_multmapfilter):
freq = svinfo.freq
chrom = svinfo.chrom
start = svinfo.start
end = svinfo.end
trans_chr, trans_start, trans_end = svinfo.trans_chrom, svinfo.trans_start, svinfo.trans_end
trans_info = "%s:%s-%s" % (trans_chr, trans_start, trans_end)
info = "%s:%s-%s" % (chrom, start, end)
svtype = "trans_chrom"
if is_single:
pass
else:
reads_type1_left, reads_type1_right, reads_type2_left, reads_type2_right, reads_type3_left, reads_type3_right, reads_type4, reads_type5_left, reads_type5_right, filtered_read_num = pos_type_classify(
bam,
chrom,
start,
end,
is_single,
read_length,
tempDir + ".info",
center=False,
maxsize=insertSize[1],
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
# exit()
trans_reads_type1_left, trans_reads_type1_right, trans_reads_type2_left, trans_reads_type2_right, trans_reads_type3_left, trans_reads_type3_right, trans_reads_type4, trans_reads_type5_left, trans_reads_type5_right, filtered_read_num = pos_type_classify(
bam,
trans_chr,
trans_start,
trans_end,
is_single,
read_length,
tempDir + ".trans",
center=False,
maxsize=insertSize[1],
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
type_reads_num = len(reads_type1_left) + len(reads_type1_right) + len(
reads_type2_left) + len(reads_type2_right) + len(
reads_type3_left) + len(reads_type3_right) + len(
reads_type4) + len(reads_type5_left) + len(
reads_type5_right)
freq = freq * (type_reads_num * 2 +
filtered_read_num) / (type_reads_num * 2)
if freq > 1:
freq = 1
modified_reads = []
modify_reads_type1 = deal_type1(ref,
reads_type1_left,
reads_type1_right,
freq,
start,
end,
svtype,
subPos=trans_info)
trans_modify_reads_type1 = deal_type1(ref,
trans_reads_type1_left,
trans_reads_type1_right,
freq,
trans_start,
trans_end,
svtype,
subPos=info)
modified_reads.extend(modify_reads_type1 + trans_modify_reads_type1)
modify_reads_type2 = deal_type2(ref,
reads_type2_left,
reads_type2_right,
freq,
start,
end,
svtype,
subPos=trans_info)
trans_modify_reads_type2 = deal_type2(ref,
trans_reads_type2_left,
trans_reads_type2_right,
freq,
trans_start,
trans_end,
svtype,
subPos=info)
modified_reads.extend(modify_reads_type2 + trans_modify_reads_type2)
modify_reads_type3 = deal_type3(reads_type3_left,
reads_type3_right,
freq,
insertSize,
start,
end,
svtype,
subPos=trans_info,
supple1=trans_reads_type3_left,
supple2=trans_reads_type3_right)
trans_modify_reads_type3 = deal_type3(trans_reads_type3_left,
trans_reads_type3_right,
freq,
insertSize,
trans_start,
trans_end,
svtype,
subPos=info,
supple1=reads_type3_left,
supple2=reads_type3_right)
if modify_reads_type3 is False or trans_modify_reads_type3 is False:
return False, "The reads are not satisfied to modify, see more details in log file"
modified_reads.extend(modify_reads_type3 + trans_modify_reads_type3)
modify_reads_type5 = deal_type5(ref,
reads_type5_left,
reads_type5_right,
freq,
start,
end,
svtype,
subPos=trans_info)
trans_modify_reads_type5 = deal_type5(ref,
trans_reads_type5_left,
trans_reads_type5_right,
freq,
trans_start,
trans_end,
svtype,
subPos=info)
modified_reads.extend(modify_reads_type5 + trans_modify_reads_type5)
return modified_reads, [], []
def deal_inversion(bam, ref, svinfo, is_single, readLength, tempDir,
insertSize, minmapq, is_multmapfilter):
freq = svinfo.freq
chrom = svinfo.chrom
start = svinfo.start
end = svinfo.end
svtype = "inv"
if is_single:
reads_type6_left, reads_type6_right, reads_type7, filtered_read_num = pos_type_classify(
bam,
chrom,
start,
end,
is_single,
readLength,
tempDir,
center=False,
maxsize=1,
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
type_reads_num = len(reads_type6_left) + len(reads_type6_right) + len(
reads_type7)
freq = freq * (type_reads_num + filtered_read_num) / type_reads_num
if freq > 1:
freq = 1
modify_reads_type6 = deal_type6(ref, reads_type6_left,
reads_type6_left, freq, start, end,
svtype)
return modify_reads_type6, [], []
else:
reads_type1_left, reads_type1_right, reads_type2_left, reads_type2_right, reads_type3_left, reads_type3_right, reads_type4, reads_type5_left, reads_type5_right, filtered_read_num = pos_type_classify(
bam,
chrom,
start,
end,
is_single,
readLength,
tempDir,
center=False,
maxsize=insertSize[1],
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
type_reads_num = len(reads_type1_left) + len(reads_type1_right) + len(
reads_type2_left) + len(reads_type2_right) + len(
reads_type3_left) + len(reads_type3_right) + len(
reads_type4) + len(reads_type5_left) + len(
reads_type5_right)
freq = freq * (type_reads_num * 2 +
filtered_read_num) / (type_reads_num * 2)
if freq > 1:
freq = 1
modified_reads = []
modify_reads_type1 = deal_type1(ref, reads_type1_left,
reads_type1_right, freq, start, end,
svtype)
modified_reads.extend(modify_reads_type1)
modify_reads_type2 = deal_type2(ref, reads_type2_left,
reads_type2_right, freq, start, end,
svtype)
modified_reads.extend(modify_reads_type2)
modify_reads_type3 = deal_type3(reads_type3_left,
reads_type3_right,
freq,
insertSize,
start,
end,
svtype,
supple1=reads_type4)
if modify_reads_type3 is False:
return False, "The reads are not satisfied to modify, see more details in log file."
modified_reads.extend(modify_reads_type3)
modify_reads_type5 = deal_type5(ref, reads_type5_left,
reads_type5_right, freq, start, end,
svtype)
modified_reads.extend(modify_reads_type5)
return modified_reads, [], []
def deal_deletion(bam, ref, svinfo, is_single, readLength, tempDir, insertSize,
minmapq, is_multmapfilter):
freq = svinfo.freq
chrom = svinfo.chrom
start = svinfo.start
end = svinfo.end
svtype = "del"
if is_single:
reads_type6_left, reads_type6_right, reads_type7, filtered_read_num = pos_type_classify(
bam, chrom, start, end, is_single, readLength, tempDir)
type_reads_num = len(reads_type6_left) + len(reads_type6_right) + len(
reads_type7)
freq = freq * (type_reads_num + filtered_read_num) / type_reads_num
if freq > 1:
freq = 1
modify_reads_type6 = deal_type6(ref,
reads_type6_left,
reads_type6_left,
freq,
start,
end,
svtype,
seqer=None)
delete_reads_type7 = deal_type7(reads_type7, freq, svtype)
return modify_reads_type6, delete_reads_type7, []
else:
reads_type1_left, reads_type1_right, reads_type2_left, reads_type2_right, reads_type3_left, reads_type3_right, reads_type4, reads_type5_left, reads_type5_right, filtered_read_num = pos_type_classify(
bam,
chrom,
start,
end,
is_single,
readLength,
tempDir,
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
type_reads_num = len(reads_type1_left) + len(reads_type1_right) + len(
reads_type2_left) + len(reads_type2_right) + len(
reads_type3_left) + len(reads_type3_right) + len(
reads_type4) + len(reads_type5_left) + len(
reads_type5_right)
if type_reads_num == 0:
return False, "All reads is filtered in this scope, total & filtered reads num %s" % filtered_read_num
freq = freq * (type_reads_num * 2 +
filtered_read_num) / (type_reads_num * 2)
if freq > 1:
freq = 1
modify_reads_type1 = deal_type1(ref, reads_type1_left,
reads_type1_right, freq, start, end,
svtype)
delete_reads_type2 = deal_type2(ref, reads_type2_left,
reads_type2_right, freq, start, end,
svtype)
supple1 = reads_type3_right + reads_type1_right
modify_reads_type3 = deal_type3(reads_type3_left,
reads_type3_right,
freq,
insertSize,
start,
end,
svtype,
supple1=supple1)
if modify_reads_type3 is False:
return False, "The reads are not satisfied to modify, see more details in log file"
delete_reads_type4 = deal_type4(reads_type4, freq, svtype)
delete_reads_type5 = deal_type5(ref, reads_type5_left,
reads_type5_right, freq, start, end,
svtype)
# modify, delete, add
return modify_reads_type1 + modify_reads_type3, delete_reads_type2 + delete_reads_type4 + delete_reads_type5, []
def deal_duplication(bam, ref, svinfo, is_single, readLength, tempDir,
insertSize, minmapq, is_multmapfilter):
freq = svinfo.freq
chrom = svinfo.chrom
start = svinfo.start
end = svinfo.end
dup_num = svinfo.dup_num
svtype = "dup"
if is_single:
reads_type6_left, reads_type6_right, reads_type7, filtered_read_num = pos_type_classify(
bam,
chrom,
start,
end,
is_single,
readLength,
tempDir,
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
type_reads_num = len(reads_type6_left) + len(reads_type6_right) + len(
reads_type7)
freq = freq * (type_reads_num + filtered_read_num) / type_reads_num
if freq > 1:
used_freq = 1
else:
used_freq = freq
add_reads_type6 = deal_type6(ref, reads_type6_left, reads_type6_left,
used_freq, start, end, svtype)
add_reads_type7 = deal_type7(reads_type7, freq, svtype)
return [], [], add_reads_type6 + add_reads_type7
else:
reads_type1_left, reads_type1_right, reads_type2_left, reads_type2_right, reads_type3_left, reads_type3_right, reads_type4, reads_type5_left, reads_type5_right, filtered_read_num = pos_type_classify(
bam,
chrom,
start,
end,
is_single,
readLength,
tempDir,
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
type_reads_num = len(reads_type1_left) + len(reads_type1_right) + len(
reads_type2_left) + len(reads_type2_right) + len(
reads_type3_left) + len(reads_type3_right) + len(
reads_type4) + len(reads_type5_left) + len(
reads_type5_right)
freq = freq * (type_reads_num * 2 +
filtered_read_num) / (type_reads_num * 2)
print freq
if freq > 1:
freq = 1
freq_dup = (dup_num - 1) * freq
freq_dup_part = freq_dup / 2.0
if freq_dup_part > 1:
freq_dup_part = 1
if freq_dup > 1:
used_freq = 1
else:
used_freq = freq_dup
# start_coverage = count_coverage(bam, chrom, start, start)
# end_coverage = count_coverage(bam, chrom, end, end)
add_reads = []
add_reads_type2 = deal_type2(ref, reads_type2_left, reads_type2_right,
used_freq, start, end, svtype)
add_reads.extend(add_reads_type2)
add_reads_type3 = deal_type3(reads_type3_left, reads_type3_right,
freq_dup, insertSize, start, end, svtype)
if add_reads_type3 is False:
return False, "The reads are not satisfied to modify, see more details in log file"
add_reads.extend(add_reads_type3)
add_reads_type4 = deal_type4(reads_type4,
freq_dup,
svtype,
insertSize=insertSize)
add_reads.extend(add_reads_type4)
add_reads_type5 = deal_type5(ref, reads_type5_left, reads_type5_right,
freq_dup_part, start, end, svtype)
add_reads.extend(add_reads_type5)
return [], [], add_reads
def deal_cnv(bam, ref, svinfo, is_single, readLength, tempDir, insertSize,
minmapq, is_multmapfilter):
freq = svinfo.freq
chrom = svinfo.chrom
start = svinfo.start
end = svinfo.end
svtype = "cnv"
if is_single:
reads_type6_left, reads_type6_right, reads_type7, filtered_read_num = pos_type_classify(
bam,
chrom,
start,
end,
is_single,
readLength,
tempDir,
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
type_reads_num = len(reads_type6_left) + len(reads_type6_right) + len(
reads_type7)
freq = freq * (type_reads_num + filtered_read_num) / type_reads_num
if svinfo.cnv_type == "loss":
delete_reads_type7 = deal_type7(reads_type7, freq, svtype)
return [], delete_reads_type7, []
elif svinfo.cnv_type == "gain":
add_reads_type7 = deal_type7(reads_type7, freq - 1, svtype)
return [], [], add_reads_type7
else:
reads_type1_left, reads_type1_right, reads_type2_left, reads_type2_right, reads_type3_left, reads_type3_right, reads_type4, reads_type5_left, reads_type5_right, filtered_read_num = pos_type_classify(
bam,
chrom,
start,
end,
is_single,
readLength,
tempDir,
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
type_reads_num = len(reads_type1_left) + len(reads_type1_right) + len(
reads_type2_left) + len(reads_type2_right) + len(
reads_type3_left) + len(reads_type3_right) + len(
reads_type4) + len(reads_type5_left) + len(
reads_type5_right)
if svinfo.cnv_type == "loss":
freq = freq * (type_reads_num * 2 +
filtered_read_num) / (type_reads_num * 2)
delete_reads_type4 = deal_type4(reads_type4,
freq,
svtype,
insertSize=insertSize,
cnvType="loss")
return [], delete_reads_type4, []
elif svinfo.cnv_type == "gain":
freq = (freq - 1) * (type_reads_num * 2 +
filtered_read_num) / (type_reads_num * 2)
add_reads_type4 = deal_type4(reads_type4,
freq,
svtype,
insertSize=insertSize,
cnvType="gain")
return [], [], add_reads_type4
def deal_translocation_balance(bam, ref, svinfo, is_single, readLength,
tempDir, insertSize, minmapq, is_multmapfilter):
freq = svinfo.freq
chrom = svinfo.chrom
start = svinfo.start
end = svinfo.end
trans_chr, trans_start, trans_end = svinfo.trans_chrom, svinfo.trans_start, svinfo.trans_end
trans_info = "%s:%s-%s" % (trans_chr, trans_start, trans_end)
info = "%s:%s-%s" % (chrom, start, end)
svtype = "trans_balance"
if is_single:
reads_type6_left, reads_type6_right, reads_type7, filtered_read_num = pos_type_classify(
bam,
chrom,
start,
end,
is_single,
readLength,
tempDir,
center=False,
maxsize=1,
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
trans_reads_type6_left, trans_reads_type6_right, trans_reads_type7, filtered_read_num = pos_type_classify(
bam,
trans_chr,
trans_start,
trans_end,
is_single,
readLength,
tempDir,
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
type_reads_num = len(reads_type6_left) + len(reads_type6_right) + len(
reads_type7)
if type_reads_num == 0:
return False, "All reads is filtered in this scope, total & filtered reads num %s" % filtered_read_num
freq = freq * (type_reads_num + filtered_read_num) / type_reads_num
if freq > 1:
freq = 1
modify_reads_type6 = deal_type6(ref,
reads_type6_left,
reads_type6_right,
freq,
start,
end,
svtype,
subPos=trans_info)
trans_modify_reads_type6 = deal_type6(ref,
trans_reads_type6_left,
trans_reads_type6_right,
freq,
trans_start,
trans_end,
svtype,
subPos=info)
return modify_reads_type6 + trans_modify_reads_type6, [], []
else:
reads_type1_left, reads_type1_right, reads_type2_left, reads_type2_right, reads_type3_left, reads_type3_right, reads_type4, reads_type5_left, reads_type5_right, filtered_read_num = pos_type_classify(
bam,
chrom,
start,
end,
is_single,
readLength,
tempDir + ".info",
center=False,
maxsize=insertSize[1],
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
# exit()
trans_reads_type1_left, trans_reads_type1_right, trans_reads_type2_left, trans_reads_type2_right, trans_reads_type3_left, trans_reads_type3_right, trans_reads_type4, trans_reads_type5_left, trans_reads_type5_right, filtered_read_num = pos_type_classify(
bam,
trans_chr,
trans_start,
trans_end,
is_single,
readLength,
tempDir + ".trans",
center=False,
maxsize=insertSize[1],
minmapq=minmapq,
is_multmapfilter=is_multmapfilter)
type_reads_num = len(reads_type1_left) + len(reads_type1_right) + len(
reads_type2_left) + len(reads_type2_right) + len(
reads_type3_left) + len(reads_type3_right) + len(
reads_type4) + len(reads_type5_left) + len(
reads_type5_right)
if type_reads_num == 0:
return False, "All reads is filtered in this scope, total & filtered reads num %s" % filtered_read_num
freq = freq * (type_reads_num * 2 +
filtered_read_num) / (type_reads_num * 2)
if freq > 1:
freq = 1
modified_reads = []
modify_reads_type1 = deal_type1(ref,
reads_type1_left,
reads_type1_right,
freq,
start,
end,
svtype,
subPos=trans_info)
trans_modify_reads_type1 = deal_type1(ref,
trans_reads_type1_left,
trans_reads_type1_right,
freq,
trans_start,
trans_end,
svtype,
subPos=info)
modified_reads.extend(modify_reads_type1 + trans_modify_reads_type1)
modify_reads_type2 = deal_type2(ref,
reads_type2_left,
reads_type2_right,
freq,
start,
end,
svtype,
subPos=trans_info)
trans_modify_reads_type2 = deal_type2(ref,
trans_reads_type2_left,
trans_reads_type2_right,
freq,
trans_start,
trans_end,
svtype,
subPos=info)
modified_reads.extend(modify_reads_type2 + trans_modify_reads_type2)
res = deal_type3(reads_type3_left,
reads_type3_right,
freq,
insertSize,
start,
end,
svtype,
subPos=trans_info,
supple1=trans_reads_type3_left,
supple2=trans_reads_type3_right)
res_trans = deal_type3(trans_reads_type3_left,
trans_reads_type3_right,
freq,
insertSize,
trans_start,
trans_end,
svtype,
subPos=info,
supple1=reads_type3_left,
supple2=reads_type3_right)
if res is False or res_trans is False:
return False, "The reads are not satisfied to modify, see more details in log file"
modify_reads_type3, add_reads_type3, delete_reads_type3 = res
trans_modify_reads_type3, trans_add_reads_type3, trans_delete_reads_type3 = res_trans
modified_reads.extend(modify_reads_type3 + trans_modify_reads_type3)
modify_reads_type5 = deal_type5(ref,
reads_type5_left,
reads_type5_right,
freq,
start,
end,
svtype,
subPos=trans_info)
trans_modify_reads_type5 = deal_type5(ref,
trans_reads_type5_left,
trans_reads_type5_right,
freq,
trans_start,
trans_end,
svtype,
subPos=info)
modified_reads.extend(modify_reads_type5 + trans_modify_reads_type5)
return modified_reads, [], []
