def compute_gc_binned_metadata(Meta,Bam):
gc_welford = {}
# for each contig
for contig in Meta.user_contigs:
# for each window
for (start,end,gc_bin,wlen) in Meta.binned_loci[contig]:
gc_count=0
key = tuple2string( ( contig, wlen, gc_bin ) )
if gc_welford.get(key)==None: gc_welford[key] = Welford()
# for each alignment
for Aln in Bam.bam.fetch(reference=contig, start=start, end=end):
if Aln.reference_start == None or Aln.reference_end==None: continue
midpoint = alignment_midpoint(Aln)
# if the mapped midpoint of the alignment is within the window
if start <= midpoint <= end:
gc_count+=1
gc_welford[key].update(gc_count)
for key in gc_welford:
Meta.gc_rc[key] = gc_welford[key].mean
Meta.gc_std[key] = gc_welford[key].std
def compute_contig_metadata(Meta,Bam):
tlen_welford = {}
# for each contig
for contig in Meta.user_contigs:
## reset these variables for each contig
if tlen_welford.get(contig)==None: tlen_welford[contig] = Welford()
read_length_sum,read_count, bp_span = 0,0,0
terminator = False
##
for (start,end) in Meta.loci[contig]:
if terminator==True: break
########
for Aln in Bam.bam.fetch(reference=contig,start=start,end=end):
if Aln.is_unmapped or type(Aln.reference_length) != int: continue
midpoint=alignment_midpoint(Aln)
if start <= midpoint <= end:
read_count+=1
read_length_sum+= Aln.reference_length
if abs(Aln.template_length)<= Meta.tlen_cap:
tlen_welford[contig].update( abs(Aln.template_length) )
if read_count > Meta.n_reads:
bp_span+= Aln.reference_end - start
terminator=True
break
if terminator==False:
bp_span += (end-start)
########
if read_count!=0:
Meta.read_len[contig] = read_length_sum / float(read_count)
else:
Meta.read_len[contig]='nan'
if bp_span!=0:
Meta.doc[contig] = (Meta.read_len[contig] * read_count) / float(bp_span)
else:
Meta.doc[contig]='nan'
Meta.tlen[contig] = tlen_welford[contig].mean
Meta.tlen_std[contig] = tlen_welford[contig].std
def gcbin_coverage(AlnFile, Meta, contig, start, end, sequence): # choose window size based on the size of the region window_size = region_window_size(Meta.window_lengths, start, end) # find the mean and STD binned coverage in the SV region in relation to GC content Region = Welford() region_windows = create_windows(window_size, sequence) for window in region_windows: local_start, local_end = window[0], window[1] window_seq = sequence[local_start:local_end] # find the gc bin for the window gc_bin = get_gc_bin(seq_gc_perc(window_seq)) # count up the number of reads in the window win_start, win_end = start+local_start, start+local_end # 1 bp overlap in windows ((0,4),(4,8),(8,12)....) read_count = 0 for Aln in AlnFile.bam.fetch(reference = contig, start = win_start, end = win_end): # skip duplicates if Aln.is_duplicate and Aln.is_unmapped: continue # Some Aln.reference_start == None, which causes error try: midpoint = alignment_midpoint(Aln) except: continue if start <= midpoint <= end: read_count += 1 # calculate the fold change of read count in window to the count of the null with similar gc content key = tuple2string( (contig, window_size, gc_bin) ) try: if int(Meta.gc_rc[key]) != 0: gc_fc = read_count / Meta.gc_rc[key] else: gc_fc = 0 except KeyError: gc_fc = 0 # add the gc fold change value to list of all windows in the sv region Region.update(gc_fc) # return mean and STD for all the windows in the region return (Region.mean, Region.std)
def gcbin_coverage(AlnFile, Meta, contig, start, end):
fasta = Meta.fasta
# choose window size based on the size of the region
region_size = end - start
window_size = Meta.window_lengths[0]
if region_size >= (2 * Meta.window_lengths[1]):
window_size = Meta.window_lengths[1]
# find the mean and STD binned coverage in the SV region in relation to GC content
Region = Welford()
read_count = 0
region_windows = create_windows(window_size, start, end)
for window in region_windows:
win_start, win_end = window[0], window[1]
# find the gc bin for the window
gc_bin = get_gc_bin(
get_gc_perc(fasta, window_size, contig, win_start, win_end))
# count up the number of reads in the window
for Aln in AlnFile.bam.fetch(reference=contig,
start=win_start,
end=win_end):
# Some Aln.reference_start == None, which causes error
try:
midpoint = alignment_midpoint(Aln)
except:
continue
if start <= midpoint <= end:
read_count += 1
# calculate the fold change of read count in window to the count of the null with similar gc content
key = tuple2string((contig, window_size, gc_bin))
gc_fc = read_count / Meta.gc_rc[key]
# add the gc fold change value to list of all windows in the sv region
Region.update(gc_fc)
# return mean and STD for all the windows in the region
return (Region.mean, Region.std)
def doc_coverage(AlnFile, Meta, contig, start, end):
meta_doc = Meta.doc[contig]
# find the doc fold change for the sv region
read_length_sum = 0
region_size = end - start
for Aln in AlnFile.bam.fetch(reference=contig, start=start, end=end):
# Some Aln.reference_start == None, which causes error
try:
midpoint = alignment_midpoint(Aln)
except:
continue
if start <= midpoint <= end:
read_length_sum += Aln.reference_length
if region_size != 0:
sv_doc = read_length_sum / region_size
if meta_doc != 0:
doc_fc = sv_doc / meta_doc
return doc_fc