def main(args):
logging.basicConfig(
level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[
logging.FileHandler("../logs/report.log"),
logging.StreamHandler()
])
logging.info(args)
paths = utils.read_paths(args.paths_file)
expr = utils.read_expr(args.expr_file)
gsm = utils.read_text(args.gsm_file)
expr = expr[gsm]
if args.n_cores > 1:
seqcorrs = utils.parallel_process(
wrapper, [(path, expr) for path in paths],
n_jobs=args.n_cores) # for some reason this is slower
else:
seqcorrs = []
for path in tqdm(paths):
seqcorrs.append(get_seq_corr(path, expr))
#seqcorrs = list(map(wrapper, tqdm([(path, expr) for path in paths])))
utils.create_dir_if_not_exist(dirname(args.out_seqcorr_file))
np.savetxt(args.out_seqcorr_file, seqcorrs)
def main(args):
logging.basicConfig(
level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[
logging.FileHandler("../logs/report.log"),
logging.StreamHandler()
])
logging.info(args)
data_raw = pd.read_excel(args.in_data_path)
row_values = data_raw[['Gene symbol', 'P-value']].values
rows = utils.parallel_process(wrap, row_values, n_jobs=args.N_cores)
rows = [elem for elem in rows if elem is not None]
genes = pd.DataFrame([val for row in rows for val in row if val[0] != -1],
columns=['EntrezID', 'P-value'])
genes_unique = genes.sort_values('P-value')
genes_unique.drop_duplicates('EntrezID', inplace=True)
genes_list = genes_unique.EntrezID.tolist()
logging.info('Num. of mapped genes: {}'.format(len(genes_list)))
utils.write_gene_list(args.out_genes_path, genes_list)
logging.StreamHandler()])
logging.info(args)
paths = utils.read_paths(args.paths_file)
expr = utils.read_expr(args.expr_file)
gsm = utils.read_text(args.gsm_file)
expr = expr[gsm]
def get_abscorr(i, j, corrdata):
method = 'pearson' if not args.spearman else 'spearman'
corrmatr = corrdata.loc[[i, j]].T.corr(method)
return corrmatr.abs().groupby('ENTREZ_GENE_ID').apply(lambda x: x.max()).T.groupby('ENTREZ_GENE_ID').apply(lambda x: x.max()).values[0, 1]
def get_seq_corr(path):
if not np.all([gene in expr.index.tolist() for gene in path]):
return np.nan
return np.mean([get_abscorr(path[i], path[i + 1], expr) for i in range(len(path) - 1)])
if args.n_cores > 1:
seqcorrs = utils.parallel_process(get_seq_corr, paths, n_jobs=args.n_cores)
else:
seqcorrs = []
for path in tqdm(paths):
seqcorrs.append(get_seq_corr(path, expr))
#seqcorrs = list(map(wrapper, tqdm([(path, expr) for path in paths])))
utils.create_dir_if_not_exist(dirname(args.out_seqcorr_file))
np.savetxt(args.out_seqcorr_file, seqcorrs)
method = 'pearson' if not args.spearman else 'spearman'
corrmatr = corrdata.loc[[i, j]].T.corr(method)
return corrmatr.abs().groupby('ENTREZ_GENE_ID').apply(lambda x: x.max(
)).T.groupby('ENTREZ_GENE_ID').apply(lambda x: x.max()).values[0, 1]
def get_seq_corr(path):
if not np.all([gene in expr.index.tolist() for gene in path]):
return np.nan
return np.mean([
get_abscorr(path[i], path[i + 1], expr) for i in range(len(path) - 1)
])
def get_random_coexpr(i):
src = random.choice(srcnodes)
dest = random.choice(destnodes)
path = random.choice(list(nx.all_shortest_paths(net, src, dest)))
return get_seq_corr(path)
if args.N_cores > 1:
seqs = utils.parallel_process(get_random_coexpr,
range(args.N_samples),
n_jobs=args.N_cores)
else:
seqs = []
for i in trange(args.N_samples):
seqs.append(get_random_coexpr(i))
np.savetxt(args.out_seqcorr_file, seqs)
parser.add_argument('--N_per_bin', type=int, default=30, help='Minimum number of nodes per degree bin')
parser.add_argument('--alpha', type=int, default=1, help='Alpha parameter of DIAMOnD')
parser.add_argument('--rdmseed', type=int, default=None, help='RNG seed')
args = parser.parse_args()
logging.basicConfig(level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[logging.FileHandler("../logs/report.log"),logging.StreamHandler()])
logging.info(args)
genes = utils.read_gene_list(args.in_genes_path)
mod = utils.read_gene_list(args.in_module_path)
net = utils.read_network(args.net_path)
rdmseeds = gen.gen_degree_preserved_sets(genes, net, args.N_samples, args.N_per_bin,rdmseed=args.rdmseed)
def wrap(pack):
rdmset, N_mod, net, alpha = pack
return rdmset + [gene[0] for gene in DIAMOnD.DIAMOnD(net, rdmset, N_mod - len(rdmset), alpha,progress_bar=False)]
samples = utils.parallel_process(wrap, utils.pack_variables(rdmseeds, len(mod), net, args.alpha), n_jobs=args.N_cores)
utils.write_genesets_list(args.out_samples_path, samples)
fcnodes = flows[(flows.FCS >= args.fc_thresh)
& (flows.N_paths >= args.npath_thresh)].index.tolist()
logging.info('Num of FC nodes: {}'.format(len(fcnodes)))
seqsims = []
for srcnodes_sample, destnodes_sample in zip(srcnodes_samples,
destnodes_samples):
all_paths = []
for src_gene in tqdm(srcnodes_sample):
def get_sps(dest_gene):
return list(nx.all_shortest_paths(net, src_gene, dest_gene))
all_paths = all_paths + sum(
utils.parallel_process(
get_sps, destnodes_sample, n_jobs=args.N_cores), [])
logging.info('Num of all paths: {}'.format(len(all_paths)))
fc_paths = []
for i in trange(len(all_paths)):
fullpath = all_paths[i]
if len(fullpath) > 2:
path = all_paths[i][1:-1]
if np.all([node in fcnodes for node in path]):
fc_paths.append(fullpath)
logging.info('Num of FC paths: {}'.format(len(fc_paths)))
go = obo_parser.GODag(args.obo_file)
gene2go = read_ncbi_gene2go(args.gene2go_file, taxids=[9606])
for j_go_term in j_go:
def wrap(i_go_term):
return resnik_sim(i_go_term, j_go_term, go, termcounts)
simlist = [sim for sim in map(wrap, i_go) if sim is not None]
if len(simlist):
sims.append(max(simlist))
return np.mean(sims)
def get_path_sim(path):
if np.all([gene in gene2go.keys() for gene in path]):
sims = []
for i in range(len(path) - 1):
sims.append(get_sim([path[i], path[i + 1]]))
return np.mean(sims)
else:
return np.nan
if args.n_cores > 1:
sims = utils.parallel_process(get_path_sim, paths, n_jobs=args.n_cores)
else:
sims = []
for path in paths:
sims.append(get_path_sim(path))
utils.create_dir_if_not_exist(dirname(args.out_sims_file))
np.savetxt(args.out_sims_file, sims)
destnodes = utils.read_gene_list(args.destnodes_file)
flows = utils.read_flows(args.flows_file)
fcnodes = flows[(flows.FCS >= args.fc_thresh)
& (flows.N_paths >= args.npath_thresh)].index.tolist()
logging.info('Num of FC nodes: {}'.format(len(fcnodes)))
all_paths = []
for src_gene in tqdm(srcnodes):
def get_sps(dest_gene):
return list(nx.all_shortest_paths(net, src_gene, dest_gene))
all_paths = all_paths + sum(
utils.parallel_process(get_sps, destnodes, n_jobs=args.N_cores), [])
logging.info('Num of all paths: {}'.format(len(all_paths)))
fc_paths = []
for i in trange(len(all_paths)):
fullpath = all_paths[i]
if len(fullpath) > 2:
path = all_paths[i][1:-1]
if np.all([node in fcnodes for node in path]):
fc_paths.append(fullpath)
logging.info('Num of FC paths: {}'.format(len(fc_paths)))
rdm_paths_A = []
for i in trange(args.N_samples):