def getRawAlignments(fn):
'''fn is the filename
returns An iterator over raw M4Records'''
itemIterator = lambda f : lineRecordIterator(f, M4Record, M4RecordTypes)
return fileIterator(fn, itemIterator)
if not len(sys.argv) > 1:
sys.exit(getHelpStr(description, arguments) + "\n")
(p_arg_map, args_remaining) = parseArgs(sys.argv[1:], arguments)
if not len(args_remaining) == 2:
sys.exit(getHelpStr(description, arguments) + "\n")
igetter = attrgetter("pctsimilarity", "qseqlength")
mpl.rc('font', family='normal', weight="normal", size=10)
mpl.rc('xtick', labelsize=5)
mpl.rc('ytick', labelsize=5)
with open(args_remaining[1]) as fh:
records = lineRecordIterator(fh, M4Record, M4RecordTypes)
items = map(igetter, records)
(pctids, lens) = zip(*items)
pp = PdfPages("out.pdf")
(fig, axScatter) = plt.subplots(figsize=(8.5, 5.5))
axScatter.scatter(lens, pctids, color="blue", s=0.001)
axScatter.set_xticks(range(0, 80000, 2000))
axScatter.set_yticks(range(0, 100, 5))
fig.suptitle(args_remaining[0])
plt.xlabel("Read Length (bp)")
plt.ylabel("Alignment Identity (%)")
divider = make_axes_locatable(axScatter)
axHistx = divider.append_axes("top", 1.2, pad=0.1, sharex=axScatter)
argument_list = [["span","span", argflag, False, "Only alignments that span the region"]]
arguments = map(CLArgument._make, argument_list)
(p_arg_map, args_remaining) = parseArgs(sys.argv[1:], arguments)
if not len(args_remaining) == 2:
sys.exit(getHelpStr(description,arguments) + "\n")
inm4 = args_remaining[0]
(chrom, rest) = args_remaining[1].split(":")
(start,end) = map(int,rest.split("-"))
it = lambda fh : lineRecordIterator(fh, M4Record, M4RecordTypes)
cond = lambda r : r.tname == chrom and not r.tend < start and not r.tstart > end
if p_arg_map["span"]:
cond = lambda r : r.tname == chrom and r.tstart < start and r.tend > end
filt_records = ifilter(cond,fileIterator(inm4, it))
for r in imap(recordToString, filt_records):
print r
def getRawAlignments(fn):
'''fn is the filename
returns An iterator over raw M4Records'''
itemIterator = lambda f: lineRecordIterator(f, M4Record, M4RecordTypes)
return fileIterator(fn, itemIterator)
args = map(CLArgument._make, argument_list)
(parg_map, args_remaining) = parseArgs(sys.argv[1:],args)
if not len(args_remaining) >= 3:
sys.exit(getHelpStr(description,args) + "\n")
title = args_remaining[0]
igetter = attrgetter("qseqlength")
flens = []
for fname in args_remaining[1:]:
with open(fname) as fh:
if parg_map["lenfiles"] < 0:
records = lineRecordIterator(fh, M4Record, M4RecordTypes)
flens.append( map(igetter, records) )
else:
getter = itemgetter(parg_map["lenfiles"])
flens.append( map(int, imap(getter, lineItemIterator(fh) )))
pp = PdfPages("hist.pdf")
fig = plt.figure()
fig.suptitle(args_remaining[0])
len_binsize = 250
plt.xlabel("Read Length")
plt.ylabel("Frequency")
if parg_map["ymax"] > 0:
plt.ylim(1,parg_map["ymax"])
from args import parseArgs, getHelpStr, argflag, CLArgument
description = ("Usage: m4region.py [options] input.m4 chr:start-end\n"
"Returns alignments that touch a region\n")
argument_list = [[
"span", "span", argflag, False, "Only alignments that span the region"
]]
arguments = map(CLArgument._make, argument_list)
(p_arg_map, args_remaining) = parseArgs(sys.argv[1:], arguments)
if not len(args_remaining) == 2:
sys.exit(getHelpStr(description, arguments) + "\n")
inm4 = args_remaining[0]
(chrom, rest) = args_remaining[1].split(":")
(start, end) = map(int, rest.split("-"))
it = lambda fh: lineRecordIterator(fh, M4Record, M4RecordTypes)
cond = lambda r: r.tname == chrom and not r.tend < start and not r.tstart > end
if p_arg_map["span"]:
cond = lambda r: r.tname == chrom and r.tstart < start and r.tend > end
filt_records = ifilter(cond, fileIterator(inm4, it))
for r in imap(recordToString, filt_records):
print r