graph[node] = {adj.split(':')[0] for adj in adjs.split(',')}
# Load connected components
CCs = []
with open('../connect_pgraph/out/pconnect1.txt') as file:
for line in file:
_, nodes = line.rstrip().split(':')
CCs.append(set(nodes.split(',')))
OGs = []
CCtypes = [{} for _ in range(5)]
for CC in CCs:
subgraph = {node: graph[node] for node in CC}
# Cluster by triangle criterion
subOGs = get_triangle_clusters(subgraph)
OGs.append(subOGs)
# Classify CCs
subnOGs = [{node for edge in subOG for node in edge} for subOG in subOGs]
if len(subnOGs) == 0:
CCtypes[0][len(subnOGs)] = CCtypes[0].get(len(subnOGs), 0) + 1 # Component has no OGs
elif len(subnOGs) == 1:
if len(subnOGs[0]) == len(CC):
CCtypes[1][len(subnOGs)] = CCtypes[1].get(len(subnOGs), 0) + 1 # Component and OG are equal
else:
CCtypes[2][len(subnOGs)] = CCtypes[2].get(len(subnOGs), 0) + 1 # Component has single OG which is a subset of the component
elif any([set.intersection(nOG1, nOG2) for nOG1, nOG2 in combinations(subnOGs, 2)]):
CCtypes[4][len(subnOGs)] = CCtypes[4].get(len(subnOGs), 0) + 1 # Component has multiple non-disjoint OGs
else:
CCtypes[3][len(subnOGs)] = CCtypes[3].get(len(subnOGs), 0) + 1 # Component has multiple pairwise disjoint OGs
rpgraph = {}
for qppid, sppids in pgraph.items():
for sppid in sppids:
try:
r = qppid in pgraph[sppid]
except KeyError:
r = False
if r:
try:
rpgraph[qppid].add(sppid)
except KeyError:
rpgraph[qppid] = {sppid}
# Cluster by triangle criterion
OGs = get_triangle_clusters(rpgraph)
# Make output directory
if not os.path.exists('out/'):
os.mkdir('out/')
# Write clusters to file
with open('out/pclusters.txt', 'w') as outfile:
for i, OG in enumerate(OGs):
OGid = hex(i)[2:].zfill(4)
outfile.write(OGid + ':' +
'\t'.join([f'{node1},{node2}'
for node1, node2 in OG]) + '\n')
"""
DEPENDENCIES
../hsps2pgraph/hsps2pgraph.py