def read_table(filename, optfile=None):
data = defaultdict(dict)
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.strip()
if not line or line.startswith('#'): continue
tokens = line.split('\t')
chrom, pos = tokens[:2]
alt = tokens[4].split(',')[0]
info = tokens[7]
# Split AC and AN from INFO field
fields = info.split(';')
fields.sort()
ac = an = None
for field in fields:
if field.startswith('AC='):
ac = int(field.split('=')[1])
if an is not None:
break
elif field.startswith('AN='):
an = int(field.split('=')[1])
if ac is not None:
break
assert ac is not None and an is not None, \
"Error: entry with AC and AN: %s" % line
data[chrom.lstrip('chr')][(int(pos), alt)] = float(ac) / an
if optfile and not os.path.isfile(optfile):
print >>sys.stderr, "Saving optimized table to:", optfile
with maybe_gzip_open(optfile, 'wb') as ofp:
cPickle.dump(data, ofp, cPickle.HIGHEST_PROTOCOL)
return data
def read_table(filename, optfile=None):
data = defaultdict(dict)
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.strip()
if not line or line.startswith('#'): continue
tokens = line.split('\t')
chrom, pos = tokens[:2]
alt = tokens[4].split(',')[0]
info = tokens[7]
# Split AC and AN from INFO field
fields = info.split(';')
fields.sort()
ac = an = None
for field in fields:
if field.startswith('AC='):
ac = int(field.split('=')[1])
if an is not None:
break
elif field.startswith('AN='):
an = int(field.split('=')[1])
if ac is not None:
break
assert ac is not None and an is not None, \
"Error: entry with AC and AN: %s" % line
data[chrom.lstrip('chr')][(int(pos), alt)] = float(ac) / an
if optfile and not os.path.isfile(optfile):
print >> sys.stderr, "Saving optimized table to:", optfile
with maybe_gzip_open(optfile, 'wb') as ofp:
cPickle.dump(data, ofp, cPickle.HIGHEST_PROTOCOL)
return data
def read_table(filename, optfile=None):
data = defaultdict(dict)
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.strip()
if not line or line.startswith('#'): continue
value, chrom, pos = line.split()
data[chrom.lstrip('chr')][int(pos)] = float(value)
if optfile and not os.path.isfile(optfile):
print >>sys.stderr, "Saving optimized table to:", optfile
with maybe_gzip_open(optfile, 'wb') as ofp:
cPickle.dump(data, ofp, cPickle.HIGHEST_PROTOCOL)
return data
def read_table(filename, optfile=None):
data = defaultdict(dict)
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.strip()
if not line or line.startswith('#'): continue
value, chrom, pos = line.split()
data[chrom.lstrip('chr')][int(pos)] = float(value)
if optfile and not os.path.isfile(optfile):
print >> sys.stderr, "Saving optimized table to:", optfile
with maybe_gzip_open(optfile, 'wb') as ofp:
cPickle.dump(data, ofp, cPickle.HIGHEST_PROTOCOL)
return data
def script(filename, quiet=False, verbose=False, **kwargs):
fields = ['MES', 'dMES', 'MES+', 'MES-', 'MEC-MC?', 'MEC-CS?', 'MES-KM?']
print '#%s' % '\t'.join(fields)
NULL = [None] * len(fields)
seqs = []
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.strip()
if line:
seqs.append(Seq(line))
sites = {}
scores = {}
# Accumulate sites for each side
for side in [3, 5]:
sites[side] = set()
for s in seqs:
sites[side].update(s.iter_seqs(side))
# Score ALL sites at once!
scores[side] = score_sites(side, sites[side])
# Print stats for each object, given queried scores
for s in seqs:
# Compute field-wise max of rows for 3' and 5'
max_row = imap(max, s.score(3, scores[3]), s.score(5, scores[5]))
print_row(max_row)
def get_genes(gene_filename=None,
cache_filename=None,
genome_filename=None,
**kwargs):
"""Loads (potentially cached) dict: gene_name -> set(genes)
If not cached, genome_filename FASTA expected to provide sequence data
"""
assert gene_filename and genome_filename or cache_filename
if cache_filename is not None and os.path.isfile(cache_filename):
print >> sys.stderr, "Loading genes from pickled file: %s" % cache_filename
with maybe_gzip_open(cache_filename) as ifp:
genes = cPickle.load(ifp)
else:
genome = Genome(genome_filename)
genes = defaultdict(set)
missed_chroms = set()
n_zero_len = 0
for entry in iter_ucsc_genes(gene_filename):
chrom = entry['chrom']
if not chrom.startswith('chr'):
chrom = 'chr%s' % chrom
if chrom not in genome:
if chrom not in missed_chroms:
print >>sys.stderr, "Could not find sequence for %s" \
" in: %s" % (chrom, genome_filename)
missed_chroms.add(chrom)
continue
# Substitute id with gene name
entry['seq'] = genome[chrom]
try:
t = Transcript(**entry)
except AssertionError, e:
if "Zero-length CDS" in str(e):
n_zero_len += 1
else:
print >>sys.stderr, "Skipping transcript: %s: %s" \
% (entry['gene'], e)
continue
if t.valid():
genes[entry['gene']].add(t)
if n_zero_len:
print >>sys.stderr, "Skipped %d transcripts with zero-length CDS" \
" annotations" % n_zero_len
if missed_chroms:
print >>sys.stderr, "Missing sequences with gene annotations: %s" \
% ', '.join(sorted(missed_chroms))
genes = dict(genes) # remove defaultdict
if cache_filename:
print >> sys.stderr, "Saving genes to pickled file: %s" % cache_filename
with open(cache_filename, 'wb') as ofp:
cPickle.dump(genes, ofp, cPickle.HIGHEST_PROTOCOL)
def read_examples(filename):
lines = []
header = []
ncols = None
with maybe_gzip_open(filename) as ifp:
header = ifp.readline().strip()
assert header.startswith('#')
header = header.replace('#', '')
for line in ifp:
line = line.strip()
if not line: continue
assert not line.startswith('#')
tokens = [float(val) for val in line.split()]
if ncols is None:
ncols = len(tokens)
else:
assert ncols == len(tokens), \
"Found row in %s with %d columns (%d expected)" % (filename, len(tokens), ncols)
lines.append(tokens)
try:
data = array(lines, dtype=float)
except ValueError:
print >> sys.stderr
# Find class column
cols = header.split()
if cols[0] == 'class':
class_col = 0
else:
class_col = None
return header, class_col, data
def read_examples(filename):
lines = []
header = []
ncols = None
with maybe_gzip_open(filename) as ifp:
header = ifp.readline().strip()
assert header.startswith('#')
header = header.replace('#', '')
for line in ifp:
line = line.strip()
if not line: continue
assert not line.startswith('#')
tokens = [float(val) for val in line.split()]
if ncols is None:
ncols = len(tokens)
else:
assert ncols == len(tokens), \
"Found row in %s with %d columns (%d expected)" % (filename, len(tokens), ncols)
lines.append(tokens)
try:
data = array(lines, dtype=float)
except ValueError:
print >>sys.stderr
# Find class column
cols = header.split()
if cols[0] == 'class':
class_col = 0
else:
class_col = None
return header, class_col, data
def iter_sequences(filename, domain=None, **kwargs):
def get_mut_seqs(seq):
pre, post = seq.split('/')
pre, old = pre.split('[')
new, post = post.split(']')
if domain:
pre_len = min(len(pre), domain)
post_len = min(len(post), domain)
# If too close to one end of sequence, accomodate
if pre_len < domain:
post_len = min(len(post), 2 * domain - pre_len)
if post_len < domain:
pre_len = min(len(pre), 2 * domain - post_len)
pre = pre[-pre_len:]
post = post[:post_len]
assert len(pre) + len(post) == 2 * domain
return pre + old + post, pre + new + post
with maybe_gzip_open(filename) as ifp:
for line in ifp:
seq = line.strip().upper()
try:
premrna = seq.replace('|', '')
postmrna = ''.join(seq.split('|')[::2])
yield get_mut_seqs(premrna), get_mut_seqs(postmrna)
except (ValueError, AssertionError):
print >> sys.stderr, "Error, invalid sequence: %s" % seq
yield None
def annotate_variants(genes, filename):
fields = ['chrom', 'pos', 'id', 'ref', 'alt', 'gene', 'tx', 'strand',
'codon', 'frame', 'premrna']
print '#%s' % '\t'.join(fields)
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.rstrip()
if not line or line.startswith('#'): continue
tokens = line.split()
chrom, pos, id, ref, alt, gene_id, tx_id = tokens[:7]
chrom = chrom[3:] if chrom.startswith('chr') else chrom
pos = int(pos)
tx = get_transcript(genes, pos, ref, alt, gene_id, tx_id)
if not tx:
logging.warning('Transcript not found for variant: %s' % line)
continue
# Get codon, frame, and mrna
cds_offset = tx.project_to_cds(pos)
aa_pos = int(cds_offset / 3) + 1
codon = tx.get_codon(aa_pos)
frame = cds_offset % 3
mut_str = tx.mutation_str(pos, ref, alt)
print '\t'.join([chrom, str(pos), id, ref, alt, tx.gene(), tx.tx(),
tx.strand(), codon, str(frame), mut_str] + tokens[7:])
def iter_sequences(filename, domain=None, **kwargs):
def get_mut_seqs(seq):
pre, post = seq.split('/')
pre, old = pre.split('[')
new, post = post.split(']')
if domain:
pre_len = min(len(pre), domain)
post_len = min(len(post), domain)
# If too close to one end of sequence, accomodate
if pre_len < domain:
post_len = min(len(post), 2*domain - pre_len)
if post_len < domain:
pre_len = min(len(pre), 2*domain - post_len)
pre = pre[-pre_len:]
post = post[:post_len]
assert len(pre) + len(post) == 2 * domain
return pre + old + post, pre + new + post
with maybe_gzip_open(filename) as ifp:
for line in ifp:
seq = line.strip().upper()
try:
premrna = seq.replace('|', '')
postmrna = ''.join(seq.split('|')[::2])
yield get_mut_seqs(premrna), get_mut_seqs(postmrna)
except (ValueError, AssertionError):
print >>sys.stderr, "Error, invalid sequence: %s" % seq
yield None
def annotate_variants(genes, filename):
fields = ['chrom', 'pos', 'id', 'ref', 'alt', 'gene', 'tx', 'strand',
'codon', 'frame', 'premrna']
print '#%s' % '\t'.join(fields)
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.rstrip()
if not line or line.startswith('#'): continue
tokens = line.split()
chrom, pos, id, ref, alt, gene_id, tx_id = tokens[:7]
chrom = chrom[3:] if chrom.startswith('chr') else chrom
pos = int(pos)
tx = get_transcript(genes, pos, ref, alt, gene_id, tx_id)
if not tx:
continue
# Get codon, frame, and mrna
cds_offset = tx.project_to_cds(pos)
aa_pos = int(cds_offset / 3) + 1
codon = tx.get_codon(aa_pos)
frame = cds_offset % 3
mut_str = tx.mutation_str(pos, ref, alt)
print '\t'.join([chrom, str(pos), id, ref, alt, tx.gene(), tx.tx(),
tx.strand(), codon, str(frame), mut_str] + tokens[7:])
def script(filename, quiet=False, verbose=False, **kwargs):
fields = ['f_premrna', 'f_mrna'] #, 'splice_dist']
print '#%s' % '\t'.join(fields)
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.strip().upper()
assert line.count('/') == 1
pre, post = line.split('/')
# Trim off mutation nucs and brackets
pre = pre[:-2] # e,g, '[A'
post = post[2:] # e.g. 'C]'
pre_chunks = pre.split('|')
post_chunks = post.split('|')
# Assume mutation is in exon
premrna_f = min(len(pre), len(post)) \
/ (len(pre) + len(post) + 1)
pre_cds = ''.join(pre_chunks[::2])
post_cds = ''.join(post_chunks[::2])
mrna_f = min(len(pre_cds), len(post_cds)) \
/ (len(pre_cds) + len(post_cds) + 1)
#splice_dist = min(len(pre_chunks[-1]), len(post_chunks[0]))
print '%.4f\t%.4f' % (premrna_f, mrna_f) #, splice_dist)
def get_genes(gene_filename=None, cache_filename=None,
genome_filename=None, **kwargs):
"""Loads (potentially cached) dict: gene_name -> set(genes)
If not cached, genome_filename FASTA expected to provide sequence data
"""
assert gene_filename and genome_filename or cache_filename
if cache_filename is not None and os.path.isfile(cache_filename):
print >>sys.stderr, "Loading genes from pickled file: %s" % cache_filename
with maybe_gzip_open(cache_filename) as ifp:
genes = cPickle.load(ifp)
else:
genome = Genome(genome_filename)
genes = defaultdict(set)
missed_chroms = set()
n_zero_len = 0
for entry in iter_ucsc_genes(gene_filename):
chrom = entry['chrom']
if not chrom.startswith('chr'):
chrom = 'chr%s' % chrom
if chrom not in genome:
if chrom not in missed_chroms:
print >>sys.stderr, "Could not find sequence for %s" \
" in: %s" % (chrom, genome_filename)
missed_chroms.add(chrom)
continue
# Substitute id with gene name
entry['seq'] = genome[chrom]
try:
t = Transcript(**entry)
except AssertionError, e:
if "Zero-length CDS" in str(e):
n_zero_len += 1
else:
print >>sys.stderr, "Skipping transcript: %s: %s" \
% (entry['gene'], e)
continue
if t.valid():
genes[entry['gene']].add(t)
if n_zero_len:
print >>sys.stderr, "Skipped %d transcripts with zero-length CDS" \
" annotations" % n_zero_len
if missed_chroms:
print >>sys.stderr, "Missing sequences with gene annotations: %s" \
% ', '.join(sorted(missed_chroms))
genes = dict(genes) # remove defaultdict
if cache_filename:
print >>sys.stderr, "Saving genes to pickled file: %s" % cache_filename
with open(cache_filename, 'wb') as ofp:
cPickle.dump(genes, ofp, cPickle.HIGHEST_PROTOCOL)
def load_data(vector_filename, log=sys.stderr):
print >>log, "Loading vector data from file: %s" % vector_filename
with maybe_gzip_open(vector_filename) as ifp:
data = loadtxt(ifp, dtype=float)
# Pop solution column
solutions = data[:, 0]
data = data[:,1:]
print >>log, "Loaded data with %d examples and %d features" % data.shape
return solutions, data
def iter_sequences(filename):
seq_re = re.compile(r'([ACGT]*)\[([ACGT])/([ACGT])\]([ACGT]*)')
with maybe_gzip_open(filename) as ifp:
for line in ifp:
seq = line.strip().upper()
mut_exons = [chunk for chunk in seq.split('|') if '/' in chunk]
assert len(mut_exons) == 1
exon = mut_exons[0]
m = seq_re.search(exon)
if m:
pre, old, new, post = m.groups()
yield pre, old, new, post
else:
print >> sys.stderr, "Error, invalid sequence: %s" % seq
yield None
def iter_sequences(filename):
seq_re = re.compile(r'([ACGT]*)\[([ACGT])/([ACGT])\]([ACGT]*)')
with maybe_gzip_open(filename) as ifp:
for line in ifp:
seq = line.strip().upper()
mut_exons = [chunk for chunk in seq.split('|') if '/' in chunk]
assert len(mut_exons) == 1
exon = mut_exons[0]
m = seq_re.search(exon)
if m:
pre, old, new, post = m.groups()
yield pre, old, new, post
else:
print >>sys.stderr, "Error, invalid sequence: %s" % seq
yield None
def iter_lines(filename):
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.strip()
if not line or line.startswith('#'): continue
ref, alt, strand, codon, offset = line.split()[:5]
assert strand in set(['+', '-', '.'])
assert len(ref) == len(alt) == 1
assert len(codon) == 3
offset = int(offset)
if strand == '-':
ref = COMPLEMENT[ref]
alt = COMPLEMENT[alt]
assert codon[offset] == ref
new_codon = codon[:offset] + alt + codon[offset+1:]
yield codon, new_codon
def iter_lines(filename):
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.strip()
if not line or line.startswith('#'): continue
ref, alt, strand, codon, offset = line.split()[:5]
assert strand in set(['+', '-', '.'])
assert len(ref) == len(alt) == 1
assert len(codon) == 3
offset = int(offset)
if strand == '-':
ref = COMPLEMENT[ref]
alt = COMPLEMENT[alt]
assert codon[offset] == ref
new_codon = codon[:offset] + alt + codon[offset + 1:]
yield codon, new_codon
def random_controls(genes, filename, match_cpg=False, avoid_splice=False):
fields = ['chrom', 'pos', 'id', 'ref', 'alt', 'gene', 'tx']
print '#%s' % '\t'.join(fields)
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.rstrip()
if not line or line.startswith('#'): continue
tokens = line.split()
chrom, pos, id, ref, alt, gene_id, tx_id = tokens[:7]
chrom = chrom[3:] if chrom.startswith('chr') else chrom
pos = int(pos)
tx = get_transcript(genes, pos, ref, alt, gene_id, tx_id)
if not tx:
continue
if match_cpg:
offset = tx.project_to_premrna(pos)
pre = tx.premrna()[offset - 1:offset]
post = tx.premrna()[offset + 1:offset + 2]
tx_ref = ref
tx_alt = alt
if tx.strand() == '-':
tx_ref = ref.translate(COMPLEMENT_TAB)
tx_alt = alt.translate(COMPLEMENT_TAB)
assert tx_ref == tx.premrna()[offset]
cpg = bool((pre and pre[0] == 'C' and
(tx_ref == 'G' or tx_alt == 'G'))
or (post and post[0] == 'G' and
(tx_ref == 'C' or tx_alt == 'C')))
else:
cpg = None
cds_offset, new_ref, new_alt = \
random_synonymous_site(tx, cpg=cpg, avoid_splice=avoid_splice)
new_pos = tx.project_from_cds(cds_offset)
if tx.strand() == '-':
new_ref = COMPLEMENT[new_ref]
new_alt = COMPLEMENT[new_alt]
print '\t'.join(
[chrom,
str(new_pos), id, new_ref, new_alt, gene_id,
tx.tx()] + tokens[7:])
def random_controls(genes, filename, match_cpg=False, avoid_splice=False):
fields = ['chrom', 'pos', 'id', 'ref', 'alt', 'gene', 'tx']
print '#%s' % '\t'.join(fields)
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.rstrip()
if not line or line.startswith('#'): continue
tokens = line.split()
chrom, pos, id, ref, alt, gene_id, tx_id = tokens[:7]
chrom = chrom[3:] if chrom.startswith('chr') else chrom
pos = int(pos)
tx = get_transcript(genes, pos, ref, alt, gene_id, tx_id)
if not tx:
continue
if match_cpg:
offset = tx.project_to_premrna(pos)
pre = tx.premrna()[offset-1:offset]
post = tx.premrna()[offset+1:offset+2]
tx_ref = ref
tx_alt = alt
if tx.strand() == '-':
tx_ref = ref.translate(COMPLEMENT_TAB)
tx_alt = alt.translate(COMPLEMENT_TAB)
assert tx_ref == tx.premrna()[offset]
cpg = bool((pre and pre[0] == 'C' and
(tx_ref == 'G' or tx_alt == 'G')) or
(post and post[0] == 'G' and
(tx_ref == 'C' or tx_alt == 'C')))
else:
cpg=None
cds_offset, new_ref, new_alt = \
random_synonymous_site(tx, cpg=cpg, avoid_splice=avoid_splice)
new_pos = tx.project_from_cds(cds_offset)
if tx.strand() == '-':
new_ref = COMPLEMENT[new_ref]
new_alt = COMPLEMENT[new_alt]
print '\t'.join([chrom, str(new_pos), id, new_ref, new_alt,
gene_id, tx.tx()] + tokens[7:])
def iter_ucsc_genes(filename):
with maybe_gzip_open(filename) as ifp:
for line in ifp:
if line.startswith('#'): continue
tokens = line.strip().split()
bin, name, chrom, strand, txStart, txEnd, cdsStart, cdsEnd, \
exonCount, exonStarts, exonEnds, score, name2 = tokens[:13]
chrom = chrom[3:] if chrom.startswith('chr') else chrom
exonStarts = exonStarts.strip(',').split(',')
exonEnds = exonEnds.strip(',').split(',')
yield {'chrom': chrom,
'tx_start': txStart,
'tx_end': txEnd,
'strand': strand,
'cds_start': cdsStart,
'cds_end': cdsEnd,
'exon_starts': exonStarts,
'exon_ends': exonEnds,
'gene': name2,
'tx': name}
def iter_ucsc_genes(filename):
with maybe_gzip_open(filename) as ifp:
for line in ifp:
if line.startswith('#'): continue
tokens = line.strip().split()
bin, name, chrom, strand, txStart, txEnd, cdsStart, cdsEnd, \
exonCount, exonStarts, exonEnds, score, name2 = tokens[:13]
chrom = chrom[3:] if chrom.startswith('chr') else chrom
exonStarts = exonStarts.strip(',').split(',')
exonEnds = exonEnds.strip(',').split(',')
yield {'chrom': chrom,
'tx_start': txStart,
'tx_end': txEnd,
'strand': strand,
'cds_start': cdsStart,
'cds_end': cdsEnd,
'exon_starts': exonStarts,
'exon_ends': exonEnds,
'gene': name2,
'tx': name}
line = line.strip()
if not line or line.startswith('#'): continue
value, chrom, pos = line.split()
data[chrom.lstrip('chr')][int(pos)] = float(value)
if optfile and not os.path.isfile(optfile):
print >> sys.stderr, "Saving optimized table to:", optfile
with maybe_gzip_open(optfile, 'wb') as ofp:
cPickle.dump(data, ofp, cPickle.HIGHEST_PROTOCOL)
return data
if optfile and os.path.isfile(optfile):
print >> sys.stderr, "Loading optimized table from:", optfile
with maybe_gzip_open(optfile, 'rb') as ifp:
table = cPickle.load(ifp)
else:
print >> sys.stderr, "Loading table from:", tablefile
table = read_table(tablefile, optfile)
print '#GERP++'
for line in sys.stdin:
line = line.strip()
if not line or line.startswith('#'): continue
chrom, pos = line.split(None)
try:
value = '%.4f' % table[chrom.lstrip('chr')][int(pos)]
except (IndexError, KeyError):
value = 'na'
def filter_variants(genes, filename, protein_coords=False):
# Do chromosomal binning to efficiently lookup overlapping transcripts
n_bins = 2048
tx_locations = defaultdict(lambda: defaultdict(list))
for gene, txs in genes.iteritems():
for tx in txs:
assert tx.gene() == gene
start = tx._cds_start + 1
end = tx._cds_end
i = int(start / n_bins)
j = int(end / n_bins)
for bin in xrange(i, j+1):
tx_locations[tx.chrom()][bin].append((start, end, tx))
def find_overlapping_transcripts(chrom, pos):
bin = int(pos / n_bins)
txs = tx_locations[chrom][bin]
return [tx for (start, end, tx) in txs if start <= pos <= end]
fields = ['chrom', 'pos', 'id', 'ref', 'alt', 'gene', 'tx']
print '#%s' % '\t'.join(fields)
n_total = 0
n_kept = 0
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.rstrip()
if not line or line.startswith('#'): continue
n_total += 1
tokens = line.split()
if protein_coords:
gene, codon, aa, mut = tokens[:4]
rest = tokens[1:]
match = get_transcript_from_protein(genes, gene, codon,
aa, mut)
if match is None:
tx = None
else:
(tx, chrom, pos, ref, alt) = match
id = '.'
else:
chrom, pos, id, ref, alts = tokens[:5]
rest = tokens[5:]
chrom = chrom[3:] if chrom.startswith('chr') else chrom
alt = alts.split(',')[0]
# Only process SNVs
if len(ref) != 1 or len(alt) != 1:
continue
pos = int(pos)
txs = []
for tx in find_overlapping_transcripts(chrom, pos):
try:
if tx.is_synonymous(pos, ref, alt):
txs.append(tx)
except AssertionError:
continue
tx = max(txs) if txs else None # Take longest valid transcript
if not tx:
continue
n_kept += 1
print '\t'.join([chrom, str(pos), id, ref, alt, tx.gene(), tx.tx()] + rest)
print >>sys.stderr, "Found %d synonymous variants (%d dropped)" % \
(n_kept, n_total - n_kept)
def filter_variants(genes, filename, protein_coords=False):
# Do chromosomal binning to efficiently lookup overlapping transcripts
n_bins = 2048
tx_locations = defaultdict(lambda: defaultdict(list))
for gene, txs in genes.iteritems():
for tx in txs:
assert tx.gene() == gene
start = tx._cds_start + 1
end = tx._cds_end
i = int(start / n_bins)
j = int(end / n_bins)
for bin in xrange(i, j+1):
tx_locations[tx.chrom()][bin].append((start, end, tx))
def find_overlapping_transcripts(chrom, pos):
bin = int(pos / n_bins)
txs = tx_locations[chrom][bin]
return [tx for (start, end, tx) in txs if start <= pos <= end]
fields = ['chrom', 'pos', 'id', 'ref', 'alt', 'gene', 'tx']
print '#%s' % '\t'.join(fields)
n_total = 0
n_kept = 0
with maybe_gzip_open(filename) as ifp:
for line in ifp:
line = line.rstrip()
if not line or line.startswith('#'): continue
n_total += 1
tokens = line.split()
if protein_coords:
gene, codon, aa, mut = tokens[:4]
rest = tokens[1:]
match = get_transcript_from_protein(genes, gene, codon,
aa, mut)
if match is None:
tx = None
else:
(tx, chrom, pos, ref, alt) = match
id = '.'
else:
chrom, pos, id, ref, alts = tokens[:5]
rest = tokens[5:]
chrom = chrom[3:] if chrom.startswith('chr') else chrom
alt = alts.split(',')[0]
# Only process SNVs
ref = ref.strip()
alt = alt.strip()
if len(ref) != 1 or len(alt) != 1:
logging.debug('Dropping non-SNP line: %s' % line)
continue
pos = int(pos)
txs = []
for tx in find_overlapping_transcripts(chrom, pos):
try:
if tx.is_synonymous(pos, ref, alt):
txs.append(tx)
except AssertionError:
continue
tx = max(txs) if txs else None # Take longest valid transcript
if not tx:
logging.debug('Variant is not synonymous on any transcript: %s' % line)
continue
n_kept += 1
print '\t'.join([chrom, str(pos), id, ref, alt, tx.gene(), tx.tx()] + rest)
logging.info("Found %d synonymous variants (%d dropped)" % \
(n_kept, n_total - n_kept))
line = line.strip()
if not line or line.startswith('#'): continue
value, chrom, pos = line.split()
data[chrom.lstrip('chr')][int(pos)] = float(value)
if optfile and not os.path.isfile(optfile):
print >>sys.stderr, "Saving optimized table to:", optfile
with maybe_gzip_open(optfile, 'wb') as ofp:
cPickle.dump(data, ofp, cPickle.HIGHEST_PROTOCOL)
return data
if optfile and os.path.isfile(optfile):
print >>sys.stderr, "Loading optimized table from:", optfile
with maybe_gzip_open(optfile, 'rb') as ifp:
table = cPickle.load(ifp)
else:
print >>sys.stderr, "Loading table from:", tablefile
table = read_table(tablefile, optfile)
print '#GERP++'
for line in sys.stdin:
line = line.strip()
if not line or line.startswith('#'): continue
chrom, pos = line.split(None)
try:
value = '%.4f' % table[chrom.lstrip('chr')][int(pos)]
except (IndexError, KeyError):
value = 'na'
