def calibrate_db(num_seqs, num_mutants, outfile, database):
db = ContigDB.load_from_filepath(database)
click.echo(f'K: {db.ramifier.k}', err=True)
prek = int(db.ramifier.k * 1.1)
contigs = random.sample(db.get_all_contigs(), num_seqs)
contigs = [
db.py_get_seq(contig_name, start_coord, start_coord + prek + 100)
for contig_name, _, start_coord, end_coord in contigs
]
contigs = [
select_one_kmer(seq, prek) for seq in contigs if len(seq) > prek
]
click.echo(f'Total contigs: {len(contigs)}', err=True)
mutated = [
mutate_seq(seq, db.ramifier.k) for seq in contigs
for _ in range(num_mutants)
]
contigs = [select_one_kmer(kmer, db.ramifier.k)
for kmer in contigs] + mutated
click.echo(f'Comparisons: {(len(contigs) ** 2) / 2 - len(contigs)}',
err=True)
dist_tbl = pd.DataFrame(py_needle(contigs),
columns=['k1', 'k2', 'levenshtein'])
def ram_dist(row):
r1, r2 = db.ramifier.ramify(row['k1']), db.ramifier.ramify(row['k2'])
return np.abs(r1 - r2).sum()
dist_tbl['ram'] = dist_tbl.apply(ram_dist, axis=1)
dist_tbl.to_csv(outfile)
def test_needle(self):
kmers = [KMER_31, MIS, GAP]
needle = py_needle(kmers, normalize=False)
for k1, k2, dist in needle:
ex_dist = 1
if k1 == GAP or k2 == GAP:
ex_dist = 2
self.assertEqual(dist, ex_dist)
def cli_lev_dist_matrix(gap, kmer_len, outfile, fasta):
kmers = parse_seqs(fasta, kmer_len, gap)
click.echo(f'{len(kmers)} unique kmers.', err=True)
start = time()
dist_tbl = pd.DataFrame(py_needle(kmers), columns=['k1', 'k2', 'lev'])
elapsed = time() - start
click.echo(f'{elapsed:.5}s to build distance matrix.', err=True)
dist_tbl.to_csv(outfile)
def calibrate_db(dropout, gap, burst, kmer_len, outfile, rotation, fasta):
seqs = [str(el.seq) for el in SeqIO.parse(fasta, 'fasta')]
kmers = set()
for seq in seqs:
for i in range(0, len(seq) - kmer_len, gap):
for j in range(burst):
j = 0
if random.random() < dropout:
kmer = seq[i + j:i + j + kmer_len]
# kmer = 'A' + kmer + 'C'
kmers.add(kmer)
# frac = 30
# mut_kmer = kmer[:(kmer_len // frac)]
# mut_kmer += mutate_seq(kmer[(kmer_len // frac):((frac - 1) * kmer_len // frac)])
# mut_kmer += kmer[((frac - 1) * kmer_len // frac):]
# kmers.add(mut_kmer)
click.echo(f'{len(kmers)} kmers', err=True)
dist_tbl = pd.DataFrame(py_needle(list(kmers)),
columns=['k1', 'k2', 'f_lev'])
if rotation is None:
ramifier = Ramifier(kmer_len)
else:
ramifier = RotatingRamifier.from_file(rotation)
def rc_lev(row):
s1, s2 = row['k1'], reverseComplement(row['k2'])
return py_needle([s1, s2])[0][2]
dist_tbl['rc_lev'] = dist_tbl.apply(rc_lev, axis=1)
dist_tbl['lev'] = dist_tbl.apply(
lambda row: min(row['f_lev'], row['rc_lev']), axis=1)
def ram_dist(row):
r1, r2 = ramifier.ramify(row['k1']), ramifier.ramify(row['k2'])
return np.abs(r1 - r2).sum()
dist_tbl['ram'] = dist_tbl.apply(ram_dist, axis=1)
dist_tbl.to_csv(outfile)
def rc_lev(row):
s1, s2 = row['k1'], reverseComplement(row['k2'])
return py_needle([s1, s2])[0][2]
def test_needle_equal(self):
kmers = [KMER_31, KMER_31]
needle = py_needle(kmers, normalize=False)
for k1, k2, dist in needle:
self.assertEqual(dist, 0)