def __init__(self, fin_godag, num_calcs, relationships, w_e, seed, prt):
tic = timeit.default_timer()
self.godag = get_godag(fin_godag,
optional_attrs=['relationship'],
prt=prt)
tic = prt_hms(tic, 'GOATOOLS read godag')
# Needed because pysemsim not understand cygwin pathes
self.graph = graph.from_resource(splitext(fin_godag)[0])
tic = prt_hms(tic, 'pygosemsim read godag')
self.seedobj = RandomSeed32(seed)
self.goids = self._init_goids(num_calcs)
tic = timeit.default_timer()
self.wang = SsWang(self.goids, self.godag, relationships, w_e)
self.go2reldepth = get_go2reldepth(
{self.godag[go]
for go in self.godag}, relationships)
tic = prt_hms(tic, 'GOATOOLS wang setup')
## MAJOR DRAWBACKS OF CURRENT METHODS ARE:
from pygosemsim import download
from pygosemsim import graph
from pygosemsim import similarity
import numpy as np
import csv
import networkx as nx
import sys
# get go database
#download.download("go-basic", "http://current.geneontology.org/ontology/go.obo")
# load go graph from downloaded file
#G = graph.from_resource("/usr/local/lib/python3.9/site-packages/pygosemsim/_resources/go-basic")
G = graph.from_resource("/root/snpXplorer/AnnotateMe/INPUTS_OTHER/20220510_go")
term_list = list(G)
# take precalculated lower bounds
similarity.precalc_lower_bounds(G)
# calculate semantic similarity -- this is an example
# similarity.lin(G, "GO:0004340", "GO:0019158")
# read file with all go terms and corresponding pvalues -- take only the significant ones
go_list = []
go_p = {}
fname = sys.argv[1]
with open(fname) as finp:
for line in finp:
line = line.rstrip().split()
def main():
#download.obo("go-basic")
#download.gaf("sgd")
G = nx.read_gpickle(gpath)
nodes = sorted(G.nodes())
node_ix = dict(zip(nodes, range(len(nodes))))
G = graph.from_resource("go-basic")
df = pd.read_csv(myers2006path, sep="\t")
df['namespace'] = [G.nodes[n]['namespace'] if n in G else None for n in df['GO ID']]
print(df)
ix = (df['namespace'] == 'biological_process') & (df['# of S. cerevisiae annotations (direct and indirect)'] > 3)
df = df[ix]
eligible_terms = set(df['GO ID'])
print("%d eligible terms" % len(eligible_terms))
sgd_to_locus = feature_preprocessing.pairwise_go_semsim.read_names("../data-sources/yeast/names.txt")
annot = annotation.from_resource("sgd")
keys = list(annot.keys())
translated_keys = {}
nodes_to_terms = {}
for key in keys:
if key not in sgd_to_locus:
continue
locus = sgd_to_locus[key]
unified = get_unified_name(locus)
if unified in node_ix:
translated_keys[key] = node_ix[unified]
node_terms = [t for t in annot[key]["annotation"].keys()]
nodes_to_terms[key] = set(node_terms) & eligible_terms
keys = list(translated_keys.keys())
print(len(keys))
#print(np.max([len(s) for s in nodes_to_terms.values()]))
F = np.zeros((len(keys), len(keys)))
for i in range(len(keys)):
a = keys[i]
print("%6d %s" % (i, a))
terms_a = nodes_to_terms[a]
if len(terms_a) == 0:
continue
for j in range(i+1, len(keys)):
b = keys[j]
terms_b = nodes_to_terms[b]
if len(terms_b) == 0:
continue
F[translated_keys[a], translated_keys[b]] = len(terms_a & terms_b)
F[translated_keys[b], translated_keys[a]] = F[translated_keys[a], translated_keys[b]]
print(np.min(F), np.max(F))
output_path = "../generated-data/pairwise_features/%s_common_functions" % (os.path.basename(gpath))
np.save(output_path, F)
import os
from pygosemsim import graph
os.chdir('/Users/hastingj/Work/Onto/addiction-ontology')
import networkx as nx
G = graph.from_resource("/Users/hastingj/Work/Onto/addiction-ontology/addicto")
nx.ancestors(G, "ADDICTO:0000212")
"ADDICTO:0000239" # Flavoured E-liquid
"ADDICTO:0000201" # Cigarette
"ADDICTO:0000212" # E-Cigarette
"ADDICTO:0000240" # Fruit flavoured e-liquid
from pygosemsim import similarity
similarity.precalc_lower_bounds(G)
# How similar are Cigarette and E-cigarette?
similarity.resnik(G, "ADDICTO:0000201", "ADDICTO:0000212")
# 2.87
similarity.wang(G, "ADDICTO:0000201", "ADDICTO:0000212")
#0.436
similarity.lin(G, "ADDICTO:0000201", "ADDICTO:0000212")
#0.45
similarity.pekar(G, "ADDICTO:0000201", "ADDICTO:0000212")
#0.455
def setUpClass(cls):
cls.G = graph.from_resource("go-basic")
graph.precalc_descendants(cls.G)
cls.annot = annotation.from_resource("goa_human")
def setUpClass(cls):
cls.G = graph.from_resource("goslim_chembl")
graph.precalc_descendants(cls.G)
def __init__(self, fin_godag, prt):
self.godag = GODag(fin_godag, optional_attrs=['relationship'], prt=prt)
self.graph = graph.from_resource(splitext(fin_godag)[0])
def main(term_type):
#download.obo("go-basic")
#download.gaf("sgd")
annot = annotation.from_resource("sgd")
G = graph.from_resource("go-basic")
#print(G.nodes['GO:2001311'])
#exit()
similarity.precalc_lower_bounds(G)
sf = functools.partial(term_set.sim_func, G, similarity.lin)
sgd_to_locus = read_names("../data-sources/yeast/names.txt")
keys = list(annot.keys())
ppcG = nx.read_gpickle(gpath)
nodes = sorted(ppcG.nodes())
node_ix = dict(zip(nodes, range(len(nodes))))
translated_keys = {}
nodes_to_terms = {}
for key in keys:
if key not in sgd_to_locus:
continue
locus = sgd_to_locus[key]
unified = get_unified_name(locus)
if unified in node_ix:
translated_keys[key] = node_ix[unified]
node_terms = [t for t in annot[key]["annotation"].keys() if G.nodes[t]['namespace'] == term_type]
nodes_to_terms[key] = set(node_terms)
keys = list(translated_keys.keys())
print(len(keys))
#print(nodes_to_terms)
#exit()
F = np.zeros((len(keys), len(keys)))
for i in range(len(keys)):
a = keys[i]
print("%6d %s" % (i, a))
terms_a = nodes_to_terms[a]
if len(terms_a) == 0:
continue
#print(terms_a)
for j in range(i+1, len(keys)):
b = keys[j]
terms_b = nodes_to_terms[b]
if len(terms_b) == 0:
continue
F[translated_keys[a], translated_keys[b]] = term_set.sim_bma(terms_a, terms_b, sf)
F[translated_keys[b], translated_keys[a]] = F[translated_keys[a], translated_keys[b]]
print(np.min(F), np.max(F))
output_path = "../generated-data/pairwise_features/%s_semsim_%s" % (os.path.basename(gpath), term_type)
np.save(output_path, F)