def kmer_homology(self, k=10, span=100):
"""Number of shared k-mers within "span" distance on either side of vertex positions"""
seq1 = ''.join([a.capitalize() for a in hg.interval(self.v1.chrom, max(1,self.v1.pos - span), min(self.v1.pos + span, hg.chrLen[hg.chrNum(self.v1.chrom)]), self.v1.strand).sequence()])
seq2 = ''.join([a.capitalize() for a in hg.interval(self.v2.chrom, max(1,self.v2.pos - span), min(self.v2.pos + span, hg.chrLen[hg.chrNum(self.v2.chrom)]), -1 * self.v2.strand).sequence()])
kset1 = Set([seq1[i:i+10] for i in range(len(seq1) - k + 1)])
kset2 = Set([seq2[i:i+10] for i in range(len(seq2) - k + 1)])
return len(kset1.intersection(kset2))
def read_cycle_file(cycle_file):
input = open(cycle_file)
segment_map = {}
interval_map = {}
cycle_map = {}
for line in input:
res = line.strip().split('\t')
if res[0] == 'Interval':
interval_map[res[1]] = hg19.interval(
res[2] if res[2].find('chr') != -1 else "chr%s" % res[2],
int(res[3]),
int(res[4]),
info={'line': res})
if res[0] == 'Segment':
segment_map[res[1]] = hg19.interval(
res[2] if res[2].find('chr') != -1 else "chr%s" % res[2],
int(res[3]),
int(res[4]),
info={'line': res})
if res[0].find('Cycle') != -1:
segments = res[0].split(';')
cycle_name = segments[0].split('=')[-1]
copy_count = float(segments[1].split('=')[-1])
cycle = segments[2].split('=')[-1].split(',')
#Sometimes a source/sink cycle will not start/end with 0, but have the 0 within the cycle
#Rotate the result such that 0s always start/end the cycle in those cases
if cycle[0][0] != '0' and len([c
for c in cycle if c[0] == '0']) > 0:
idx = cycle.index([c for c in cycle if c == '0+'][0])
cycle = cycle[idx:] + cycle[:idx]
cycle_map[cycle_name] = {'cycle': cycle, 'copy_count': copy_count}
input.close()
return (segment_map, interval_map, cycle_map)
def __init__(self, segment_list=None, cycle_list=None, ilist=None, file=None, file_content=None):
if file is not None or file_content is not None:
self.segment_list = hg.interval_list([])
self.segment_dict = {}
self.cycle_dict = {}
self.ilist = hg.interval_list([])
if file_content:
lines = file_content.split('\n')
else:
lines = str(open(file).read().decode()).split('\n')
ll = [l.strip().split() for l in lines if len(l.strip()) > 0]
for l in ll:
if 'Segment' == l[0]:
s = hg.interval(l[2], int(l[3]), int(l[4]), info=[l[1]])
self.segment_dict[l[1]] = s
self.segment_list.append(s)
elif 'Cycle=' in l[0]:
ls = l[0].split(';')
ci = ls[0].split('=')[1]
cn = float(ls[1].split('=')[1])
cl = []
for s in ls[2].split('=')[1].split(','):
if s[-1] == '+':
cl.append((s[:-1], 1))
else:
cl.append((s[:-1], -1))
self.cycle_dict[ci] = (ci, cn, cl)
elif 'Interval' == l[0]:
self.ilist.append(hg.interval(l[2], int(l[3]), int(l[4]), info=[l[1]]))
elif cycle_list is None:
segment_set = hg.interval_list([hg.interval(ss[0], ss[1], ss[2]) for ss in {(s.chrom, s.start, s.end) for s in segment_list}])
segment_set.sort()
self.segment_list = segment_set
self.segment_dict = {}
seg_id = {}
cl = []
for s in enumerate(segment_set):
self.segment_dict[str(s[0] + 1)] = s[1]
seg_id[(s[1].chrom, s[1].start, s[1].end)] = str(s[0] + 1)
for s in segment_list:
cl.append((seg_id[(s.chrom, s.start, s.end)], s.strand))
for ii in range(len(self.segment_list)):
s = self.segment_list[ii]
s.info = [seg_id[(s.chrom, s.start, s.end)]]
self.cycle_dict = {'1':('1', 1, cl)}
self.ilist = hg.interval_list([s[0] for s in segment_set.merge_clusters(extend=1)])
for ii in range(len(self.ilist)):
self.ilist[ii].info = [str(ii)]
else:
self.segment_list = segment_list
self.segment_dict = {s.info[0]: s for s in segment_list}
self.cycle_dict = {c[0]:c for c in cycle_list}
if ilist is not None:
self.ilist = ilist
else:
self.ilist = hg.interval_list([s[0] for s in segment_list.merge_clusters(extend=1)])
for ii in range(len(self.ilist)):
self.ilist[ii].info = [str(ii)]
def load_bed(bed_file, value = None, log = False, sep='\t'):
bed_data = hg19.interval_list()
for line in open(bed_file):
res = line.split(sep)
if value is None:
bed_data.append(hg19.interval(res[0], int(res[1]), int(res[2]), info={'value':float(res[3]) if not log else 10**float(res[3])}))
else:
bed_data.append(hg19.interval(res[0], int(res[1]), int(res[2]), info={'value':value}))
bed_data.sort()
return bed_data
def pivot(self, c1, si1, si2):
cycle1 = self.cycle_dict[c1]
# check if segments overlap
if not self.segment_dict[cycle1[2][si1][0]].intersects(
self.segment_dict[cycle1[2][si2][0]]):
raise Exception("Segments do not overlap")
# check if segments have opposite orientation
if cycle1[2][si1][1] == cycle1[2][si2][1]:
raise Exception("Segments should be in opposite orientation")
seg1 = self.segment_dict[cycle1[2][si1][0]]
seg2 = self.segment_dict[cycle1[2][si2][0]]
seg1_found = False
seg2_found = False
for i in self.segment_list:
if (i.chrom, i.start, i.end) == (seg1.chrom, seg1.start, seg2.end):
seg1_found = True
ns1 = i.info[0]
overlap1 = (ns1, cycle1[2][si1][1])
if (i.chrom, i.start, i.end) == (seg1.chrom, seg2.start, seg1.end):
seg2_found = True
ns2 = i.info[0]
overlap2 = (ns2, cycle1[2][si2][1])
if not seg1_found:
ns1 = self.next_seg_id()
overlap1 = (ns1, cycle1[2][si1][1])
self.segment_dict[ns1] = hg.interval(seg1.chrom,
seg1.start,
seg2.end,
info=[ns1])
self.segment_list.append(self.segment_dict[ns1])
if not seg2_found:
ns2 = self.next_seg_id()
overlap2 = (ns2, cycle1[2][si2][1])
self.segment_dict[ns2] = hg.interval(seg1.chrom,
seg2.start,
seg1.end,
info=[ns2])
self.segment_list.append(self.segment_dict[ns2])
cycle1_init = cycle1[2][:si1]
if cycle1[2][si1][1] == -1:
(overlap1, overlap2, ns1, ns2) = ((overlap2[0], -1 * overlap2[1]),
(overlap1[0],
-1 * overlap1[1]), ns2, ns1)
cycle1_span = [(s[0], -1 * s[1]) for s in cycle1[2][si1 + 1:si2][::-1]]
cycle1_final = cycle1[2][si2 + 1:]
mcycle = cycle1_init + [overlap1] + cycle1_span + [overlap2
] + cycle1_final
mcycle_id = self.next_cycle_id()
self.cycle_dict[mcycle_id] = (mcycle_id, cycle1[1], mcycle)
self.cycle_dict[c1] = (c1, 0.0, cycle1[2])
return
def parse_bed(segment_file):
input = open(segment_file, 'r')
amplicons = hg19.interval_list()
for line in input:
res = line.strip().split('\t')
amplicons.append(hg19.interval(res[0], int(res[1]), int(res[2])))
return amplicons
def parse_cycles_file(cycles_file, addchr = False):
cycles = {}
segSeqD = {}
with open(cycles_file) as infile:
for line in infile:
if line.startswith("Segment"):
fields = line.rstrip().split()
lowerBound = int(fields[3])
upperBound = int(fields[4])
chrom = fields[2]
if addchr:
chrom = "chr%s" % chrom
segNum = fields[1]
segSeqD[segNum] = hg19.interval(chrom,lowerBound,upperBound, info={'name':segNum})
elif "Cycle=" in line:
curr_cycle = []
fields = line.rstrip().rsplit(";")
lineD = {x.rsplit("=")[0]:x.rsplit("=")[1] for x in fields}
segs = lineD["Segments"].rsplit(",")
#TODO: Need to rotate the segs in case there's a 0 within the path, or else path is incorrect
skip = False
for i in segs:
seg = i[:-1]
if seg != "0":
strand = i[-1]
curr_cycle.append((seg,strand))
else:
skip = True
if not skip:
cycles[lineD["Cycle"]] = curr_cycle
return cycles,segSeqD
def build_segments(self, bed_data = None):
if bed_data is None:
bed_data = self.bed_data
points_x = []
points_y = []
colors = []
fpoints_x = []
fpoints_y = []
fcolors = []
previous_end = total_length_with_spacing*(global_rot/360.0)
for ind,sp in enumerate(start_points):
start_point = int(previous_end - sp)
start_angle = start_point/total_length_with_spacing*360
end_angle = (start_point - lens[ind])/total_length_with_spacing*360
#segseqD referenced as global variable here because I'm lazy
segment = segSeqD[cycle[ind][0]]
strand = cycle[ind][1]
hits = [h[0] for h in bed_data.intersection([segment])]
if self.color_bed is not None:
color_subhits = hg19.interval_list([h[0] for h in self.color_bed.intersection([segment])])
for h in hits:
for pos in xrange(h.start, h.end, self.point_spacing):
if pos > segment.end or pos < segment.start:
continue
if self.color_bed is not None:
temp = hg19.interval(h.chrom, pos, pos)
color_hits = color_subhits.intersection([temp],self.point_spacing)
if len(color_hits) != 0:
color = color_hits[0][0].info['color']
else:
color = self.color if 'color' not in h.info else h.info['color']
else:
color = self.color if 'color' not in h.info else h.info['color']
if strand == "+":
normStart = start_point - max(0,pos-segment.start)
normEnd = start_point - min(segment.end-segment.start,pos-segment.start)
else:
normEnd = start_point - min(segment.end-segment.start,segment.end-pos)
normStart = start_point - max(0,segment.end - pos)
hvalue = h.info['value'] if h.info['value'] > self.ymin else self.ymin
hvalue = hvalue if hvalue < self.ymax else self.ymax
y_scale_value = (1.*hvalue-self.ymin)/(self.ymax-self.ymin)
if self.is_log:
y_scale_value = (math.log10(hvalue)-math.log10(self.ymin))/(math.log10(self.ymax)-math.log10(self.ymin))
r_scale_value = y_scale_value*(self.track_rmax-self.track_rmin)+self.track_rmin
x_s,y_s = pol2cart(r_scale_value,normStart/total_length_with_spacing*2*np.pi)
if 'fill' in h.info:
foo = fpoints_x.append(x_s)
foo = fpoints_y.append(y_s)
fcolors.append(color)
else:
foo = points_x.append(x_s)
foo = points_y.append(y_s)
colors.append(color)
foo = ax.scatter(points_x,points_y,marker='o',s=1,linewidths=0.01,facecolors='none',color=colors)
foo = ax.scatter(fpoints_x,fpoints_y,marker='*',s=1,linewidths=0.01,color=fcolors)
def kmer_homology(self, k=10, span=100):
seq1 = ''.join([
a.capitalize() for a in hg.interval(
self.v1.chrom, max(1, self.v1.pos - span),
min(self.v1.pos + span, hg.chrLen[hg.chrNum(self.v1.chrom)]),
self.v1.strand).sequence()
])
seq2 = ''.join([
a.capitalize() for a in hg.interval(
self.v2.chrom, max(1, self.v2.pos - span),
min(self.v2.pos +
span, hg.chrLen[hg.chrNum(self.v2.chrom)]), -1 *
self.v2.strand).sequence()
])
kset1 = Set([seq1[i:i + 10] for i in range(len(seq1) - k + 1)])
kset2 = Set([seq2[i:i + 10] for i in range(len(seq2) - k + 1)])
return len(kset1.intersection(kset2))
def build_genebed_from_fpkm(fpkm):
fpkm_bed = hg19.interval_list()
for (g,f) in fpkm.items():
if g not in ensembl_grc37_map:
continue
gene = ensembl_grc37_map[g]
fpkm_bed.append(hg19.interval(gene.chrom, gene.start, gene.end, info={'value':f, 'name':gene.info['Name']}))
fpkm_bed.sort()
return fpkm_bed
def load_ensembl_grc37():
input = open('/pedigree2/projects/namphuon/data/references/hg19/annotations/Homo_sapiens.GRCh37.64.gtf' ,'r')
ensemble_data = hg19.interval_list()
ensembl_grc37_map = {}
for line in input:
res = line.split('\t')
info = dict([r.strip().replace('"','').split(' ') for r in res[-1].strip().split('; ') if len(r.split(' ')) == 2])
temp = hg19.interval("chr%s" % res[0],int(res[3]),int(res[4]),info={'data':info})
ensemble_data.append(temp)
foo = ensembl_grc37_map.setdefault(info['gene_id'],[]).append(temp)
ensembl_grc37 = hg19.interval_list()
for g in ensembl_grc37_map:
start = min([e.start for e in ensembl_grc37_map[g]])
end = max([e.end for e in ensembl_grc37_map[g]])
name = [e.info['data']['gene_name'] for e in ensembl_grc37_map[g] if 'gene_name' in e.info['data']]
name = name[0] if len(name) >= 1 else g
ensembl_grc37.append(hg19.interval("%s" % e.chrom, start, end, info={'intervals':ensembl_grc37_map[g],'GeneID':g,'Name':name}))
input.close()
ensembl_grc37.sort()
for e in ensembl_grc37:
ensembl_grc37_map[e.info['GeneID']] = e
return (ensembl_grc37, ensembl_grc37_map)
def sequence(self, flank_size=-1):
if self.edge_type == 'sequence':
seq = hg.interval(self.v1.chrom, self.v1.pos,
self.v2.pos).sequence()
if flank_size > 0:
seq = hg.interval(self.v1.chrom, self.v1.pos - flank_size + 1,
self.v1.pos).sequence() + seq + hg.interval(
self.v2.chrom, self.v2.pos,
self.v2.pos + flank_size - 1).sequence()
else:
if self.hom == None:
seq = 'N' * 20
else:
seq = self.hom_seq
if flank_size == -1:
flank_size = 1000
if flank_size > 0:
if self.hom is not None and self.hom > 0:
hom = self.hom
else:
hom = 0
if self.edge_type == 'source':
if self.v2.strand == -1:
right_seq = hg.interval(
self.v2.chrom, self.v2.pos + hom,
self.v2.pos + hom + flank_size - 1).sequence()
left_seq = ''
else:
left_seq = hg.interval(
self.v2.chrom, self.v2.pos - hom - flank_size + 1,
self.v2.pos - hom).sequence()
right_seq = ''
elif self.v1.strand == 1:
left_seq = hg.interval(self.v1.chrom,
self.v1.pos - hom - flank_size + 1,
self.v1.pos - hom).sequence()
if self.v2.strand == -1:
right_seq = hg.interval(
self.v2.chrom, self.v2.pos + hom,
self.v2.pos + hom + flank_size - 1).sequence()
else:
right_seq = hg.interval(self.v2.chrom,
self.v2.pos - hom -
flank_size + 1,
self.v2.pos - hom,
strand=-1).sequence()
else:
right_seq = hg.interval(self.v1.chrom, self.v1.pos + hom,
self.v1.pos + hom + flank_size -
1).sequence()
if self.v2.strand == -1:
left_seq = hg.interval(self.v2.chrom,
self.v2.pos + hom,
self.v2.pos + hom + flank_size -
1,
strand=-1).sequence()
else:
left_seq = hg.interval(
self.v2.chrom, self.v2.pos - hom - flank_size + 1,
self.v2.pos - hom).sequence()
seq = left_seq + seq + right_seq
return seq
else:
samp_name = args.sname.rsplit("/")[-1]
fname = samp_name
bed_feat_dict = {}
if args.bed_files:
for i,j in zip(args.bed_files,args.feature_labels):
print j,i
#feature name -> chromosome -> ordered list of positions
bed_list = parse_bed_file(i)
bed_feat_dict[j] = feat_bed_to_lookup(bed_list)
outer_bar = max(bed_track_height*(len(bed_feat_dict)+2),10)
bed_data = hg19.interval_list([hg19.interval('chr8', 127638302, 127938302, info={'value':int(random.random()*100)}), hg19.interval('chr8', 128716346,128746346, info={'value':int(random.random()*100)})])
bed_data.sort()
args.prefix_name = '/pedigree2/projects/namphuon/programs/CycleViz/COLO320DM'
args.cycles_file = '/pedigree2/projects/namphuon/data/paul_gbm39/unsorted/COLO320_DM_S270/onco_amplicon1_cycles.txt'
args.fpkm_file = '/pedigree2/projects/namphuon/data/paul_gbm39/unsorted/COLO320_DM_S270/colo320dm.fpkm.csv'
args.wgs_file = '/pedigree2/projects/namphuon/data/paul_gbm39/unsorted/COLO320_DM_S270/colo320dm.wgs.1000.pileup.log.bed'
cycles_numbers = ['6', '9', '10', '12', '13', '14', '15', '16','19']
args.atac_peak_file = '/pedigree2/projects/namphuon/data/paul_gbm39/unsorted/ATAC-seq/SRC1655_summits_250ext_q1e6_nochrM_merged.bed'
args.atac_file = '/pedigree2/projects/namphuon/results/paul_gbm39/ATAC/COLO320DM.atac.1000.pileup.log.bed'
args.prefix_name = '/pedigree2/projects/namphuon/programs/CycleViz/PC3'
args.cycles_file = '/nucleus/pedigree/projects/extrachromosome/data/turner2017/reconstruction/run14/FF-77_amplicon4_cycles.txt'
args.fpkm_file = '/pedigree2/projects/namphuon/results/paul_gbm39/rnaseq/PC3.fpkm.csv'
args.wgs_file = '/pedigree2/projects/namphuon/results/paul_gbm39/PC3/PC3.wgs.1000.pileup.log.bed'
a.intersection(ci[1]).size()
for ci in hg.interval_list([a]).intersection(cr)
])) or a.size() - sum([
a.intersection(ci[1]).size()
for ci in hg.interval_list([a]).intersection(cr)
]) > 2000000):
if (len(hg.interval_list([a]).intersection(cr))) == 0:
uc_list.append(a)
else:
cra = hg.interval_list([a]).intersection(cr)
cpos = a.start
for crai in cra:
if cpos < crai[1].start - 1000000:
uc_list.append(
hg.interval(a.chrom,
cpos,
crai[1].start - 1000000,
info=a.info))
cpos = crai[1].end + 1000000
if a.end > cpos:
uc_list.append(hg.interval(a.chrom, cpos, a.end, info=a.info))
uc_list = hg.interval_list([
a for a in uc_list if float(a.info[-1]) *
a.segdup_uniqueness() > GAIN and a.rep_content() < 2.5
])
uc_merge = uc_list.merge_clusters(extend=300000)
with open(outname, "w") as outfile:
for a in uc_merge:
if sum([ai.size() for ai in a[1]]) > CNSIZE_MIN:
outfile.write('\t'.join([
from collections import defaultdict
import pysam
import hg19util as hg
f = pysam.AlignmentFile("/pedigree2/projects/namphuon/data/SCC090/pacbio/merged.bam")
segs = defaultdict(lambda: [], {})
readlen = {}
refi = hg.interval_list([hg.interval(i) for i in f.references])
segi = 1
qi = 0
qindex = {}
qlist = []
for l in f.fetch():
ref = l.reference_name.split(':')[0]
ref_start = int(l.reference_name.split(':')[1].split('-')[0]) + l.reference_start
ref_end = int(l.reference_name.split(':')[1].split('-')[0]) + l.reference_end
qstart = l.query_alignment_start
qend = l.query_alignment_end
if l.query_name not in qindex:
qindex[l.query_name] = qi
qlist.append(l.query_name)
qi += 1
if l.is_reverse:
qstart = l.infer_query_length() - l.query_alignment_end
qend = l.infer_query_length() - l.query_alignment_start
for a in bamFile:
vlist[(a.qname, a.is_read1)].append(a)
if a.tid == -1 or bamFile.getrname(a.tid) not in hg19refs:
if a.tid == -1:
continue
vreads[bamFile.getrname(a.tid)].add((a.qname, a.is_read1))
continue
if caln is not None and (a.pos > caln.pos + 300 or caln.tid != a.tid
) and clean_genomic_cluster(clist):
clusterList.append(clist)
if caln is None or a.pos > caln.pos + 300 or caln.tid != a.tid:
clist = []
caln = a
# if hg.interval(a, bamfile=bamFile).num_unmasked() >= 35:
if hg.interval(a, bamfile=bamFile).rep_content() <= 3 and a.mapq >= 10:
clist.append(a)
if caln is not None and (a.pos > caln.pos + 300 or
caln.tid != a.tid) and clean_genomic_cluster(clist):
clusterList.append(clist)
# clusterList.sort(key=cmp_to_key(lambda x, y: hg.interval(bamFile.getrname(x[0].tid), x[0].pos, x[-1].pos + x[-1].infer_query_length()) > hg.interval(bamFile.getrname(y[0].tid), y[0].pos, y[-1].pos + y[-1].infer_query_length())))
clusterList.sort(
key=lambda x: hg.interval(bamFile.getrname(x[0].tid), x[0].pos, x[-1].pos +
x[-1].infer_query_length()),
reverse=True)
vsuper = {
v: set([
v2 for v2 in vreads if v2 != v and len(vreads[v]) < len(vreads[v2])
and vreads[v].issubset(vreads[v2])
a.intersection(ci[1]).size()
for ci in hg.interval_list([a]).intersection(cr)
])) or a.size() - sum([
a.intersection(ci[1]).size()
for ci in hg.interval_list([a]).intersection(cr)
]) > 2000000):
if (len(hg.interval_list([a]).intersection(cr))) == 0:
uc_list.append(a)
else:
cra = hg.interval_list([a]).intersection(cr)
cpos = a.start
for crai in cra:
if cpos < crai[1].start - 1000000:
uc_list.append(
hg.interval(a.chrom,
cpos,
crai[1].start - 1000000,
info=a.info))
cpos = crai[1].end + 1000000
if a.end > cpos:
uc_list.append(
hg.interval(a.chrom, cpos, a.end, info=a.info))
uc_list = hg.interval_list([
a for a in uc_list if float(a.info[1]) *
a.segdup_uniqueness() > 5.0 and a.rep_content() < 2.5
])
uc_merge = uc_list.merge_clusters(extend=300000)
all_uc = hg.interval_list([
a[0] for a in uc_merge if sum([ai.size() for ai in a[1]]) > CNSIZE_MIN
])
for a in bamFile:
vlist[(a.qname, a.is_read1)].append(a)
if a.tid == -1 or bamFile.getrname(a.tid) not in hg19refs:
if a.tid == -1:
continue
vreads[bamFile.getrname(a.tid)].add((a.qname, a.is_read1))
continue
if caln is not None and (a.pos > caln.pos + 300 or caln.tid != a.tid
) and clean_genomic_cluster(clist):
clusterList.append(clist)
if caln is None or a.pos > caln.pos + 300 or caln.tid != a.tid:
clist = []
caln = a
# if hg.interval(a, bamfile=bamFile).num_unmasked() >= 35:
if hg.interval(a, bamfile=bamFile).rep_content() <= 3 and a.mapq >= 10:
clist.append(a)
if caln is not None and (a.pos > caln.pos + 300 or
caln.tid != a.tid) and clean_genomic_cluster(clist):
clusterList.append(clist)
clusterList.sort(lambda x, y: hg.interval(
bamFile.getrname(x[0].tid), x[0].pos, x[-1].pos + x[-1].infer_query_length(
)) > hg.interval(bamFile.getrname(y[0].tid), y[0].pos, y[-1].pos + y[-1].
infer_query_length()))
vsuper = {
v: Set([
v2 for v2 in vreads if v2 != v and len(vreads[v]) < len(vreads[v2])
and vreads[v].issubset(vreads[v2])
])
parser = argparse.\
ArgumentParser(description="Cycles File")
parser.add_argument('--cycles',
dest='cycles_file',
help="File listing cycles in amplicon",
metavar='FILE',
action='store',
type=str,
nargs=1)
args = parser.parse_args()
cycles_file = args.cycles_file[0]
ll = [l.strip().split() for l in open(cycles_file) if len(l.strip()) > 0]
segments = hg.interval_list([
hg.interval(l[2], int(l[3]), int(l[4]), info=[int(l[1])]) for l in ll
if l[0] == 'Segment'
])
for s in segments:
if s.chrom[:3] == 'chr':
s.info.append('Human')
else:
s.info.append('Viral')
segments.sort()
segment_id_dict = {s.info[0]: s for s in segments}
cycles = []
for c in [l[0].split(';') for l in ll if 'Cycle=' in l[0]]:
c_dict = {cc.split('=')[0]: cc.split('=')[1] for cc in c}
new_dict = {}
new_dict['Cycle'] = int(c_dict['Cycle'])
def merge(self, c1, c2, si1, si2):
cycle1 = self.cycle_dict[c1]
cycle2 = self.cycle_dict[c2]
# check if atmost 1 cycle has source vertex
if '0' in [s[0] for s in cycle1[2]] and '0' in [s[0] for s in cycle2[2]]:
raise Exception("Cannot merge 2 cycles with source vertices")
# if cycle2 has source vertex, exchange c1,c2
if '0' in [s[0] for s in cycle2[2]]:
(c1, c2, si1, si2, cycle1, cycle2) = (c2, c1, si2, si1, cycle2, cycle1)
if si1 == 0 or si1 == len(cycle1[2]) - 1:
raise Exception("Cannot use source segment for merging")
# check if segments overlap
if not self.segment_dict[cycle1[2][si1][0]].intersects(self.segment_dict[cycle2[2][si2][0]]):
raise Exception("Segments do not overlap" + str(self.segment_dict[cycle1[2][si1][0]]) + " " + str(self.segment_dict[cycle2[2][si2][0]]))
# cnlist: (merged cn, cycle1cn, cycle2cn)
if cycle1[1] == 0 or cycle2[1] == 0:
raise Exception("Cycle copy numbers should be > 0 to merge")
if cycle1[1] > cycle2[1]:
cnlist = (cycle2[1], cycle1[1] - cycle2[1], 0.0)
else:
cnlist = (cycle1[1], 0.0, cycle2[1] - cycle1[1])
seg1 = self.segment_dict[cycle1[2][si1][0]]
seg2 = self.segment_dict[cycle2[2][si2][0]]
seg1_found = False
seg2_found = False
for i in self.segment_list:
if cycle1[2][si1][1] == 1 and (i.chrom, i.start, i.end) == (seg1.chrom, seg1.start, seg2.end):
seg1_found = True
ns1 = i.info[0]
overlap1 = (ns1, cycle1[2][si1][1])
elif cycle1[2][si1][1] == -1 and (i.chrom, i.start, i.end) == (seg1.chrom, seg2.start, seg1.end):
seg1_found = True
ns1 = i.info[0]
overlap1 = (ns1, cycle1[2][si1][1])
if cycle1[2][si1][1] == 1 and (i.chrom, i.start, i.end) == (seg1.chrom, seg2.start, seg1.end):
seg2_found = True
ns2 = i.info[0]
overlap2 = (ns2, cycle1[2][si1][1])
elif cycle1[2][si1][1] == -1 and (i.chrom, i.start, i.end) == (seg1.chrom, seg1.start, seg2.end):
seg2_found = True
ns2 = i.info[0]
overlap2 = (ns2, cycle1[2][si1][1])
if not seg1_found:
ns1 = self.next_seg_id()
overlap1 = (ns1, cycle1[2][si1][1])
if cycle1[2][si1][1] == 1:
self.segment_dict[ns1] = hg.interval(seg1.chrom, seg1.start, seg2.end, info=[ns1])
else:
self.segment_dict[ns1] = hg.interval(seg1.chrom, seg2.start, seg1.end, info=[ns1])
self.segment_list.append(self.segment_dict[ns1])
if not seg2_found:
ns2 = self.next_seg_id()
overlap2 = (ns2, cycle1[2][si1][1])
if cycle1[2][si1][1] == 1:
self.segment_dict[ns2] = hg.interval(seg1.chrom, seg2.start, seg1.end, info=[ns2])
else:
self.segment_dict[ns2] = hg.interval(seg1.chrom, seg1.start, seg2.end, info=[ns2])
self.segment_list.append(self.segment_dict[ns2])
cycle1_init = cycle1[2][:si1]
if not cycle1[2][si1][1]:
(overlap1, overlap2, ns1, ns2) = (overlap2, overlap1, ns2, ns1)
if cycle1[2][si1][1] == cycle2[2][si2][1]:
cycle2_span = cycle2[2][si2 + 1:] + cycle2[2][:si2]
else:
cycle2_span = [(s[0], -1 * s[1]) for s in cycle2[2][:si2][::-1] + cycle2[2][si2 + 1:][::-1]]
cycle1_final = cycle1[2][si1 + 1:]
mcycle = cycle1_init + [overlap1] + cycle2_span + [overlap2] + cycle1_final
mcycle_id = self.next_cycle_id()
self.cycle_dict[mcycle_id] = (mcycle_id, cnlist[0], mcycle)
self.cycle_dict[c1] = (c1, cnlist[1], cycle1[2])
self.cycle_dict[c2] = (c2, cnlist[2], cycle2[2])
return
if a.tid == -1:
continue
vreads[bamFile.getrname(a.tid)].add((a.qname, a.is_read1))
continue
if caln is not None and (a.pos > caln.pos + 300 or caln.tid != a.tid) and clean_genomic_cluster(clist):
clusterList.append(clist)
if caln is None or a.pos > caln.pos + 300 or caln.tid != a.tid:
clist = []
caln = a
# if hg.interval(a, bamfile=bamFile).num_unmasked() >= 35:
# if hg.interval(a, bamfile=bamFile).rep_content() <= 3 and a.mapq >= 10:
clist.append(a)
if caln is not None and (a.pos > caln.pos + 300 or caln.tid != a.tid) and clean_genomic_cluster(clist):
clusterList.append(clist)
clusterList.sort(lambda x, y: hg.interval(bamFile.getrname(x[0].tid), x[0].pos, x[-1].pos + x[-1].infer_query_length()) > hg.interval(bamFile.getrname(y[0].tid), y[0].pos, y[-1].pos + y[-1].infer_query_length()))
vsuper = {v: Set([v2 for v2 in vreads if v2 != v and len(vreads[v]) < len(vreads[v2]) and vreads[v].issubset(vreads[v2])]) for v in vreads}
vequaldict = {}
vequal = []
for v in vreads:
inserted = False
for vset in vequal:
if vreads[v] == vreads[vset[0]]:
vset[1].add
vequaldict[v] = vset
inserted = True
break
if not inserted:
vset = (v, Set([v]))
vequal.append(vset)
segments = []
# segments=hg.interval_list(rdAlts.replace('.bed', '_segments.bed'), 'bed')
# bandsfile="karyotype.HK359.EGFR.txt"
# segments = [(l[2], hg.interval(l[1], int(l[4]), int(l[5])).intersection(i), l[6]) for l in [ll.strip().split() for ll in open(bandsfile) if 'band' in ll and ll.strip().split()[1][:3] == 'chr'] if hg.interval(l[1], int(l[4]), int(l[5])).intersects(i)]
# segments = [('', hg.interval(l[1], int(l[4]), int(l[5])), l[6]) for l in [ll.strip().split() for ll in open(bandsfile) if 'band' in ll and ll.strip().split()[1][:3] == 'chr']]
if args.extendmode == 'VIRAL':
logging.info("#TIME " + '%.3f\t' % (clock() - TSTART) +
"Finding integration sites: " + str(rdList[0]))
de = bamFileb2b.interval_discordant_edges(rdList)
old_stdout = sys.stdout
sys.stdout = mystdout = StringIO()
amplist = bamFileb2b.interval_hops(rdList, explore=False)
alist = hg.interval_list(
[hg.interval(e[0].v1.chrom, e[0].v1.pos, e[0].v1.pos) for e in de] +
[hg.interval(e[0].v2.chrom, e[0].v2.pos, e[0].v2.pos)
for e in de] + rdList)
alist.sort()
rdList = hg.interval_list([
i[0] for i in alist.merge_clusters(extend=5000000) if len(
hg.interval_list([i[0]]).intersection(amplist) +
hg.interval_list([i[0]]).intersection(rdList)) > 0
])
rdList = hg.interval_list([
hg.interval(i.chrom, max(0, i.start - 10000),
min(i.end + 10000, hg.chrLen[hg.chrNum(i.chrom)]))
for i in rdList
])
iout = open(outName + '.integration_search.out', 'w')
iout.write(mystdout.getvalue())
def draw_episome(self, input_files, output_file=None, auto_scale=0):
cycles_section_top = 30
cycles_section_size = 0
space_between_decompositions = 90 / (1 + auto_scale)
bottoms = []
for i in range(len(input_files)):
if i != 0:
cycles_section_size += space_between_decompositions
input_content = input_files[i][1]
intervals, segments, seg_name_to_index_map, cycles, directions, chr_offs, copy_counts, cycles_names = self.readDataFile(
input_content)
number_of_element = sum(len(x) for x in cycles)
cycles_section_size += number_of_element * 10
for cycle in cycles:
if cycle[-1] != 0:
cycles_section_size += 10
bottoms.append(cycles_section_top + cycles_section_size)
tops = [cycles_section_top] + [
bottom + space_between_decompositions for bottom in bottoms[:-1]
]
cycles_section_bottom = cycles_section_top + cycles_section_size
# print ('cycle section_top:', cycles_section_top)
# print ('cycle section_bottom:', cycles_section_bottom)
for i in range(len(input_files)):
input_content = input_files[i][1]
self.file_names.append(
Text('%s: %s' % (str(i + 1), input_files[i][0]), 0.5,
tops[i] - 27))
# print ('bottom:', bottoms[i])
# print ('top:', tops[i])
intervals, segments, seg_name_to_index_map, cycles, directions, chr_offs, copy_counts, cycles_names = self.readDataFile(
input_content)
if i == 0:
self.reconstructed_cycles = [cname for cname in cycles_names]
self.reconstructed_segments = [
segment_count for segment_count in range(len(segments))
]
self.compute_chr_offsets(chr_offs)
if len(intervals) == 0:
intervals = self.compute_intervals(segments)
ilist = hg.interval_list([
hg.interval(chr_name, start_point, end_point)
for chr_name, start_point, end_point in intervals
])
maxIntvl = self.findMaxIntervals(segments)
sortedL = {}
compact = {}
span = {}
for ch in maxIntvl.keys():
sortedL[ch] = self.makeListOfSegmentEndPoints(segments, ch)
sortedCopy = list(sortedL[ch])
mergeL = self.mergeIntervals(sortedCopy)
compact[ch], span[ch] = self.compactIntervals(
mergeL, maxIntvl[ch])
newsegs = self.convertSegmentCoordinates(segments, span, compact,
maxIntvl, ilist)
if i == 0:
self.drawSections(intervals, ilist, cycles_section_top,
cycles_section_bottom, auto_scale)
# self.drawAxesLabels(sortedL, compact, maxIntvl, span, chr_offs, bottoms[i])
# self.drawAxes(span, compact, maxIntvl, chr_offs, bottoms[i])
# self.drawAxesDottedLines(sortedL, compact, maxIntvl, span, chr_offs, tops[i], bottoms[i])
self.drawCycles(newsegs, seg_name_to_index_map, cycles, directions,
tops[i], chr_offs, i, copy_counts, cycles_names)