Last active
August 29, 2015 14:06
-
-
Save ctb/cb8e8632b7d7de681a68 to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
#!/usr/bin/python | |
import sys | |
import screed | |
if len(sys.argv) != 4: | |
print >>sys.stderr, "USAGE: compare.py <mutations.txt> <orig.fasta> <corrected.fasta>" | |
sys.exit(1) | |
fp_out = open('fp.out', 'w') | |
fn_out = open('fn.out', 'w') | |
mutations = {} | |
orig_reads = {} | |
for line in open(sys.argv[1]): | |
lexemes = line.strip().split("\t") | |
if len(lexemes) != 4: | |
continue | |
read_name = lexemes[0] | |
pos = int(lexemes[1]) | |
orig = lexemes[2].upper() | |
mut = lexemes[3].upper() | |
if read_name not in mutations: | |
mutations[read_name] = {} | |
mutations[read_name][pos] = {"correct" : orig, "mutation" : mut} | |
for n, record in enumerate(screed.open(sys.argv[2])): | |
orig_reads[record["name"]] = record["sequence"].upper() | |
tps, fps, tns, fns, tot_errors = 0, 0, 0, 0, 0 | |
print >>sys.stderr, "#read\ttp\tfp\ttn\tfn\ttotal_errors" | |
for n, record in enumerate(screed.open(sys.argv[3])): | |
name = record["name"] | |
seq = record["sequence"][:100].upper() | |
orig = orig_reads[name] | |
read_mut = mutations.get(name, {}) | |
tp, fp, tn, fn = 0, 0, 0, 0 | |
for pos in range(len(seq)): | |
if pos >= len(orig): | |
print >>sys.stderr, "{0}\n{1}\n{2}".format(name, seq, orig) | |
if pos in read_mut: | |
if seq[pos] == read_mut[pos]["correct"]: | |
tp += 1 | |
else: | |
fn += 1 | |
print >>fn_out, seq, orig, pos | |
else: | |
if seq[pos] == orig[pos]: | |
tn += 1 | |
else: | |
fp += 1 | |
print >>fp_out, seq, orig, pos | |
tps += tp | |
fps += fp | |
tns += tn | |
fns += fn | |
tot_errors += len(read_mut) | |
print "{0}\t{1}\t{2}\t{3}\t{4}\t{5}".format(name, tp, fp, tn, fn, len(read_mut)) | |
print >>sys.stderr, "Totals\t\t{0}\t{1}\t{2}\t{3}\t{4}".format(tps, fps, tns, fns, tot_errors) |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
551 export PYTHONPATH=~/dev/khmer:~/dev/screed | |
532 python ~/dev/nullgraph/make-reads.py --mutation-details details.out genome.fa > reads.fa | |
536 ../scripts/normalize-by-median.py -k 20 -C 20 reads.fa -s reads.ht | |
539 python ../sandbox/read_aligner.py reads.ht reads.fa > out 2> errout | |
552 python ../sandbox/read_aligner.py reads.ht reads.fa > out 2> errout | |
554 ../../2013-read-aligner/compare.py details.out reads.fa out |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Foo. |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
#! /usr/bin/env python | |
import sys | |
import khmer | |
kh = khmer.load_counting_hash(sys.argv[1]) | |
print kh.ksize() | |
for line in open(sys.argv[2]): | |
line = line.strip() | |
seq_a, seq_b, mutpos = line.split() | |
mutpos = int(mutpos) | |
for pos in range(len(seq_a) - kh.ksize() + 1): | |
signal = '' | |
if pos == mutpos: | |
signal = '*' | |
print '%s%s:%s,%s' % (signal, pos, kh.get(seq_a[pos:pos+kh.ksize()]), | |
kh.get(seq_b[pos:pos+kh.ksize()]),), | |
print '' | |
print '---' |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment