fn_totalkmers = '%s_%s_totalkmers.pkl' % (name, str(k))
path_totalkmers = os.path.join(work_dir, fn_totalkmers)
if not os.path.exists(work_dir):
os.makedirs(work_dir)
if os.path.exists(fn_totalkmers):
print('%s already exists' % (fn_totalkmers))
sys.exit()
print('Calculating kmer totals for %s using model %s.' %
(path_reads, path_model))
total_kmers = r2v.calc_total_kmers(path_reads,
path_model,
k,
verbose=True,
v=v)
print('Dump total kmers to %s' % (path_totalkmers))
six.moves.cPickle.dump(total_kmers, open(path_totalkmers, 'wb'), protocol=4)
r2v.embed_reads(path_reads,
path_totalkmers,
path_model,
work_dir,
normread=nr,
k=k,
a=a,
verbose=True,
v=v)
name = 'ag'
a = 1e-05
path_sample = argv[1]
path_model = argv[2]
fn_model_base = path_model.split('/')[-1]
k = int(fn_model_base.split('_')[1])
fn_model_base = '_'.join(fn_model_base.split('_')[1:-1])
fn_out = '%s_%s_total_kmers_split.pkl' % (name, fn_model_base)
dir_totalkmers = '/mnt/HA/groups/rosenGrp/embed_samples/data/%s/total_kmers' % (
name)
path_totalkmers = os.path.join(dir_totalkmers, fn_out)
path_out = '/mnt/HA/groups/rosenGrp/embed_samples/data/%s/embeddings_split' % (
name)
path_out = os.path.join(path_out, fn_model_base)
if not os.path.exists(path_out):
os.makedirs(path_out)
r2v.embed_reads(path_sample,
path_totalkmers,
path_model,
path_out,
k=k,
a=a,
delim=' ',
verbose=True,
v=1000)
if os.path.exists(fn_totalkmers):
print('%s already exists' % (fn_totalkmers))
sys.exit()
print('Calculating kmer totals for samples in %s using model %s.' %
(samp_dir, path_model))
total_kmers = r2v.calc_total_kmers_split(samp_dir,
path_model,
k,
verbose=True,
v=v)
print('Dump total kmers to %s' % (path_totalkmers))
six.moves.cPickle.dump(total_kmers, open(path_totalkmers, 'wb'), protocol=4)
for samp in glob(samp_dir + '/*'):
print('Embedding sample %s.' % (samp))
r2v.embed_reads(samp,
path_totalkmers,
path_model,
work_dir,
k=k,
a=a,
svm=False,
normread=False,
to_sample=True,
delim=' ',
verbose=True,
v=1000)