def write_enrichment_files(in_fname, out_fname, go_dct):
'''
Takes in a cluster name, reads it, and writes to the out_fname.
'''
# Compute GO enrichment for networks without GO.
cluster_dct = file_operations.get_cluster_dictionary(in_fname)
compute_go_enrichments(out_fname, cluster_dct, go_dct)
def write_summary(clus_fname, net_fname, eval_fname, enrich_fname, out_fname):
# Get cluster dictionary. Values are lists of genes.
clst_go_dct = file_operations.get_cluster_dictionary(clus_fname)
# Get the number of gene-gene and gene-GO edges from the network.
num_genes_net, num_gg_net, num_ggo_net = file_operations.get_network_stats(
net_fname)
# Get density dictionary. Values are in-density and out-density.
density_dct = file_operations.get_cluster_densities(eval_fname)
# Find the best p-value GO enrichments for each cluster.
enrichment_dct = file_operations.get_enrichment_dct(enrich_fname)
# Write out to file.
out = open(out_fname, 'w')
out.write('num_genes_in_net\tnum_g_g_net\tnum_g_go_net\n')
out.write('%s\t%d\t%d\n' % (num_genes_net, num_gg_net, num_ggo_net))
out.write('cluster_number\tin_dens\tout_dens\tratio\t')
out.write('num_genes\tnum_go_terms_in\tnum_g_g_edges\tnum_g_go_edges\t')
out.write('top_enrichment_p\tgo_terms_in\n')
for cid in density_dct:
cluster_go_terms = []
# cid = str(i + 1)
clus = clst_go_dct[cid]
num_genes = len(clus)
num_go, num_gg, num_ggo = 0, 0, 0
for node in clus:
if ('ENSMUSG' not in node and 'ENSG' not in node):
num_go += 1
cluster_go_terms += [node]
# for edge in edge_list_go:
# node_a, node_b = edge
# if node_a in clus and node_b in clus:
# if 'ENSMUSG' not in node_a:
# # This means we have a GO term.
# num_ggo += 1
# elif 'ENSMUSG' not in node_b:
# num_ggo += 1
# else:
# # This means that we have a gene-gene edge.
# num_gg += 1
# assert(num_gg % 2 == 0)
# assert(num_ggo % 2 == 0)
# # Divide each of these values by two because each edge is written
# # twice in the network.
# num_gg /= 2
# num_ggo /= 2
in_dens, out_dens, ratio = density_dct[cid]
# ratio = in_dens / (in_dens + out_dens)
out.write('%s\t%g\t%g\t%g\t' % (cid, in_dens, out_dens, ratio))
out.write('%d\t%d\t%d\t%d\t' % (num_genes, num_go, num_gg, num_ggo))
out.write('%s\t%s\n' % (enrichment_dct[cid], '\t'.join(
cluster_go_terms)))
out.close()
def write_summary_tables(net_type):
'''
Generates the filenames for each of the files we read from to summarize
the clustering performances of a particular run.
'''
results_folder = './results/%s_results' % species
# Generate the format string.
format_str = (results_folder, n_clusters, net_type)
# Simulated annealing cluster results filename.
if net_type == 'none':
clus_fname = '%s/%s_clusters_%s.txt' % format_str
else:
clus_fname = '%s/%s_clusters_%s_clean.txt' % format_str
clus_dct = file_operations.get_cluster_dictionary(clus_fname)
# Perl script evaluation results filename.
eval_fname = '%s/%s_cluster_eval_%s.txt' % format_str
density_dct = get_cluster_densities(eval_fname)
# DBGAP enrichment results filename.
dbgap_fname = '%s/%s_clusters_dbgap_terms_%s.txt' % format_str
dbgap_dct = get_enrichment_dct(dbgap_fname)
out_fname = '%s/%s_clusters_summary_%s.tsv' % format_str
out = open(out_fname, 'w')
out.write('Cluster ID\tIn-Density\tOut-Density\tIn/(In+Out)\t'
'Gene size\tTop DBGAP p-value\tTop DBGAP term\n')
for cid in sorted(density_dct.keys(), key=lambda x: int(x)):
clus = clus_dct[cid]
num_genes = len(clus)
in_dens, out_dens = density_dct[cid]
best_dbgap_p = dbgap_dct[cid][1][0]
best_dbgap_term = dbgap_dct[cid][0][0]
out.write(
'%s\t%g\t%g\t%g\t%d\t%s\t%s\n' %
(cid, in_dens, out_dens, in_dens /
(in_dens + out_dens), num_genes, best_dbgap_p, best_dbgap_term))
out.close()
def compute_label_enrichments(in_fname, out_fname):
'''
Takes in a cluster name, reads it, and writes to the out_fname.
'''
global gene_universe
cluster_dct = file_operations.get_cluster_dictionary(in_fname)
net_genes = [item for sublist in cluster_dct.values() for item in sublist]
gene_universe = set(net_genes)
out = open(out_fname, 'w')
# Loop through the clusters.
for clus_id in cluster_dct:
clus_genes = set(cluster_dct[clus_id])
sorted_fisher_dct = get_sorted_fisher_dct(clus_genes)
# Get the log of the top 5 enrichment p-values.
top_label_terms, top_p_values = [], []
for (label_term, p_value) in sorted_fisher_dct[:5]:
top_label_terms += [label_term]
top_p_values += [str(p_value)]
out.write(
'Cluster %s\n%s\n%s\n' %
(clus_id, '\t'.join(top_label_terms), '\t'.join(top_p_values)))
out.close()