def train_vblur_from_meta(hist_fn, cds_fa, vblur_fname):
""" procedure pipeline """
cds_range = get_cds_range(cds_fa)
tlist = parse_rlen_hist(hist_fn)
b, ptrue, eps = meta_pipeline(tlist, cds_range, utr5_offset, imax,
rlen_min, rlen_max, converge_cutoff)
write_vblur(b, vblur_fname)
def deblur_pipeline(bam_fname, cds_fa, oprefix, force):
""" full pipeline for deblur ribo profiles for a given sample """
# step 0: prepare input parameters
odir = os.path.dirname(oprefix)
if odir and not os.path.exists(odir): os.makedirs(odir)
if oprefix.endswith("/"): oprefix += "ribo"
raw_hist = "{}_raw.hist".format(oprefix)
high_cov_hist = "{}_hc.hist".format(oprefix)
vblur_fname = "{}.vblur".format(oprefix)
eps_fname = "{}.eps".format(oprefix)
profile_fname = "{}.profile".format(oprefix)
# step 1: generate length-specific profiles
if not os.path.exists(raw_hist) or force == True:
group_reads_by_length(bam_fname, raw_hist)
if not os.path.exists(raw_hist) or os.path.getsize(raw_hist) == 0:
print("FATAL: deblur_pipeline failed at group_reads_by_len!", file=sys.stderr)
print("abort program!", file=sys.stderr)
exit(1)
else: print("length-specific profile exists, use cached", file=sys.stderr)
# step 2: filter high-coverage profiles
if not os.path.exists(high_cov_hist) or force == True:
filter_high_cover_profile(raw_hist, cds_fa, cover_ratio, cnt_threshold, high_cov_hist)
if not os.path.exists(high_cov_hist) or os.path.getsize(high_cov_hist) == 0:
print("FATAL: deblur_pipeline failed at filter_high_cover_profile!", file=sys.stderr)
print("abort program!", file=sys.stderr)
exit(1)
else: print("high-coverage profile exists, use cached", file=sys.stderr)
# step 3: train blur vector from meta profiles
if not os.path.exists(vblur_fname) or force == True:
train_vblur_from_meta(high_cov_hist, cds_fa, vblur_fname)
if not os.path.exists(vblur_fname) or os.path.getsize(vblur_fname) == 0:
print("FATAL: deblur_pipeline failed at train_vblur_from_meta!", file=sys.stderr)
print("abort program!", file=sys.stderr)
exit(1)
else: print("vblur file exists, use cached", file=sys.stderr)
# step 4: deblur high-coverage profiles
if not os.path.exists(eps_fname) or force == True:
deblur_transcripts(high_cov_hist, cds_fa, vblur_fname, eps_fname)
if not os.path.exists(eps_fname) or os.path.getsize(eps_fname) == 0:
print("FATAL: deblur_pipeline failed at deblur_transcripts!", file=sys.stderr)
print("abort program!", file=sys.stderr)
exit(1)
else: print("deblur file exits, use cached", file=sys.stderr)
# step 5: combine length-specific profiles
if not os.path.exists(profile_fname) or force == True:
cds_range = get_cds_range(cds_fa)
ctrue_merge = construct_deblur_profiles(eps_fname, vblur_fname, raw_hist, cds_range)
aprof = batch_build_Aprof(ctrue_merge, cds_range, -utr5_offset, asite_offset)
write_profiles(aprof, profile_fname)
if not os.path.exists(profile_fname) or os.path.getsize(profile_fname) == 0:
print("FATAL: deblur_pipeline failed at combine_profile!", file=sys.stderr)
print("abort program!", file=sys.stderr)
exit(1)
else: print("final results exists, nothing needs to be done", file=sys.stderr)
def filter_high_cover_profile(hist_fn, cds_fa, cover_ratio, cnt_threshold,
ofname):
""" pipeline for filtering high coverage profiles """
cds_range = get_cds_range(cds_fa)
tlist = parse_rlen_hist(hist_fn)
tprofile = get_transcript_profiles(tlist, cds_range, utr5_offset,
utr3_offset)
pcelebrity = filter_transcript_profiles(tprofile, cds_range, cnt_threshold,
cover_ratio)
tid2rid = {t['tid']: rid for rid, t in tlist.items()}
write_rlen_hist(pcelebrity, cds_range, tid2rid, ofname)
plt.show()
if __name__ == "__main__":
if len(sys.argv) != 6:
print "Usage: python frameshift_celebrity.py rlen.hist cds_range.txt rlen.vblur rlen.eps output_dir"
exit(1)
hist_fn = sys.argv[1]
cds_txt = sys.argv[2]
vblur_fname = sys.argv[3]
deblur_fname = sys.argv[4]
odir = sys.argv[5]
ensure_dir(odir)
fname = get_file_core(hist_fn)
print "get cds range"
cds_range = get_cds_range(cds_txt)
print "parse read len hist file"
tlist = parse_rlen_hist(hist_fn)
print "get pre-computed blur vector"
b = read_vblur(vblur_fname)
print "get pre-computed deblur results"
ptrue, eps = read_essentials(deblur_fname)
print "construct cobs all at once"
tprofile = get_transcript_profiles(tlist, cds_range, utr5_offset,
utr3_offset)
#cobs = build_cobs_with_shifts(tprofile, cds_range, utr5_offset, utr3_offset, rlen_min, rlen_max, klist)
cobs = construct_all_cobs(tprofile, cds_range, utr5_offset, utr3_offset,
rlen_min, rlen_max)
print "construct ctrue all at once"
ctrue = batch_build_ctrue(ptrue, eps, cobs)
tid_list = np.array(cobs.keys())
return means, varrs, skews
if __name__ == "__main__":
if len(sys.argv) != 7:
print "Usage: python elongation_rate.py transcript.fasta cds_range.txt rlen.hist rlen.vblur rlen.eps output_dir"
exit(1)
tfasta = sys.argv[1]
cds_txt = sys.argv[2]
hist_fn = sys.argv[3]
vblur_fname = sys.argv[4]
deblur_fname = sys.argv[5]
odir = sys.argv[6]
ensure_dir(odir)
fname = get_file_core(hist_fn)
print "get cds range"
cds_range = get_cds_range(cds_txt)
tseq = get_tseq(tfasta, cds_range)
print "parse read len hist file"
tlist = parse_rlen_hist(hist_fn)
print "get pre-computed blur vector"
b = read_vblur(vblur_fname)
print "get pre-computed deblur results"
ptrue, eps = read_essentials(deblur_fname)
print "construct cobs all at once"
tprofile = get_transcript_profiles(tlist, cds_range, utr5_offset, utr3_offset)
# cobs = build_cobs_with_shifts(tprofile, cds_range, utr5_offset, utr3_offset, rlen_min, rlen_max, klist)
cobs = construct_all_cobs(tprofile, cds_range, utr5_offset, utr3_offset, rlen_min, rlen_max)
print "construct ctrue all at once"
ctrue_rlen = batch_build_ctrue(ptrue, eps, cobs)
mprof = { tid: merge_profiles(plist) for tid, plist in ctrue_rlen.iteritems() }
base_prof = batch_build_Aprof(mprof, cds_range, -utr5_offset, asite_offset)