def make_cnl_file(self, filepath=None):
local_cnl = self.persistent_cnl()
if local_cnl:
return utils.link_file(local_cnl, destname=filepath)
else:
assert filepath is not None
return self.save_to_cnl_file(filepath)
def run(self, data, **kwargs):
if False and (data.is_directed()):
raise Exception("only undirected is supported")
params = dict(kwargs)
params = {k:v for k, v in params.items() if v is not None }
if not data.is_directed():
params['Sym'] = 1
params['d'] = "output"
if "r" not in params: params['r'] = 0.1
if not utils.file_exists(data.file_edges):
data.to_edgelist()
txt_params = " ".join(["-{} {}".format(k, v) for k, v in params.items()])
cmd = "{} -jar {} -i {} {} ".format(utils.get_java_command(), config.GANXISW_PROG, "edges.txt", txt_params)
with utils.TempDir() as tmp_dir:
utils.remove_if_file_exit(os.path.join(tmp_dir, "output"), True)
utils.create_dir_if_not_exists(os.path.join(tmp_dir, "output"))
self.logger.info("Running " + cmd)
utils.link_file(data.file_edges, tmp_dir, "edges.txt")
timecost, status = utils.timeit(lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception("Run command with error status code {}".format(status))
outputfile = glob.glob(os.path.join(tmp_dir, "output/SLPAw*.icpm"))[0]
clusters = []
with open (os.path.join(tmp_dir, outputfile), "r") as f:
for line in f:
clusters.append([int(u) for u in line.strip().split(" ")])
clusters = dict(enumerate(clusters))
self.logger.info("Made %d clusters in %f seconds" % (len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self, data, cutoff_r=0.01, inflation_in=2, NBDisimilarity_q=0.3, seed=None):
if False and (data.is_directed()):
raise Exception("only undirected is supported")
if seed is not None:self.logger.info("seed ignored")
params = locals();del params['self'];del params['data']
if not utils.file_exists(data.file_edges):
data.to_edgelist()
cmd = "{} {} {} {} {}".format(config.LABLE_RANK_PROG, "edges.txt", cutoff_r, inflation_in, NBDisimilarity_q)
with utils.TempDir() as tmp_dir:
utils.remove_if_file_exit(os.path.join(tmp_dir, "output"), True)
utils.create_dir_if_not_exists(os.path.join(tmp_dir, "output"))
self.logger.info("Running " + cmd)
utils.link_file(data.file_edges, tmp_dir, "edges.txt")
timecost, status = utils.timeit(lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception("Run command with error status code {}".format(status))
outputfile = glob.glob(os.path.join(tmp_dir, "output/LabelRank*.icpm"))[0]
clusters = []
with open (os.path.join(tmp_dir, outputfile), "r") as f:
for line in f:
clusters.append([int(u) for u in line.strip().split(" ")])
clusters = dict(enumerate(clusters))
self.logger.info("Made %d clusters in %f seconds" % (len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self, data, seed=None):
self.logger.warning(
"dct assumes node starts with zero and is continuous")
if seed is None:
seed = np.random.randint(999999)
params = {'seed': seed, 'prog': self.progname}
if (data.is_directed() or data.is_weighted()) and False:
raise Exception(
"only undirected and unweighted graph is supported")
if not utils.file_exists(data.file_edges):
data.to_edgelist()
with utils.TempDir() as tmp_dir:
#tmp_dir = "/tmp/abc"
pajek = utils.link_file(data.file_edges,
dest_dir=tmp_dir,
destname='edges.txt')
cmd = "{} {}".format(
config.get_dct_prog(self.progname, data.is_directed()), pajek)
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(lambda: utils.shell_run_and_wait(
cmd, tmp_dir, env={'SEED': str(seed)}))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
outputfiles = glob.glob(tmp_dir + "/output*.cluster")
import pandas as pd
df_from_each_file = (pd.read_csv(f, sep=" ", header=None)
for f in outputfiles)
output = pd.concat(df_from_each_file, ignore_index=True)
output.columns = ['node', 'cluster']
clusters = output[['cluster', 'node']]
clusters = clusters.groupby('cluster').apply(
lambda u: list(u['node'])).to_dict()
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self, data, seed=None):
self.logger.warning(
"dct::seq_louvain assumes node starts with zero and is continuous")
if seed is None:
seed = np.random.randint(999999)
params = {'seed': seed}
if (data.is_directed() or data.is_weighted()) and False:
raise Exception(
"only undirected and unweighted graph is supported")
if not utils.file_exists(data.file_edges):
data.to_edgelist()
with utils.TempDir() as tmp_dir:
pajek = utils.link_file(data.file_edges,
dest_dir=tmp_dir,
destname='edges.txt')
cmd = "{} -f -s {} -o output {}".format(
config.get_dct_prog('infomap', data.is_directed()), seed,
pajek)
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
outputfile = tmp_dir + "/output"
import pandas as pd
output = pd.read_csv(outputfile, sep=" ", header=None)
output.columns = ['cluster']
output['node'] = range(len(output))
clusters = output[['cluster', 'node']]
clusters = clusters.groupby('cluster').apply(
lambda u: list(u['node'])).to_dict()
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self, data, f=False, m=None,seed=None):
if False and (data.is_directed()):
raise Exception("only undirected is supported")
if seed is not None:self.logger.info("seed ignored")
params = {}
params['f'] = f
params['m'] = m
if not utils.file_exists(data.file_hig):
data.to_higformat()
cmd = ["./hirecs"]
cmd.append("-oje")
if f: cmd.append('-f')
if m is not None: cmd.append("-m{}".format(m))
cmd.append(data.file_hig)
cmd.append("> output")
cmd = " ".join(cmd)
with utils.TempDir() as tmp_dir:
with open(os.path.join(tmp_dir, "tmpcmd"), 'wt') as f: f.write(cmd)
self.logger.info("Running " + cmd)
cmd = "bash tmpcmd"
utils.link_file(os.path.join(config.HIRECS_PATH, 'hirecs'), tmp_dir)
utils.link_file(os.path.join(config.HIRECS_PATH, 'libhirecs.so'), tmp_dir)
utils.link_file(data.file_hig, tmp_dir)
timecost, status = utils.timeit(lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception("Run command with error status code {}".format(status))
with open (os.path.join(tmp_dir, "output"), "r") as f:
output = json.load(f)
mod = output['mod']
communities = output['communities']
clusters = {}
for c in communities:
clusters[int(c)] = [int(u) for u in communities[c].keys()]
self.logger.info("Made %d clusters in %f seconds with modularity %f" % (len(clusters), timecost, mod))
result = {}
result['runname'] = self.name
result['params'] = params
result['overlap'] = True
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self, data, **kwargs):
if False and (data.is_directed() or data.is_weighted()):
raise Exception(
"only undirected and unweighted graph is supported")
params = dict(kwargs)
params = {u: v for u, v in params.items() if v is not None}
if "loss" not in params: params['loss'] = 'modularity'
argparams = dict(params)
if argparams['loss'] in ['pmod', 'mom']:
if argparams['loss'] not in argparams:
raise Exception(
"You have to specify pmod=<val> or mom=<val> for the loss function"
)
loss_args = "--loss {{{},{}}}".format(argparams['loss'],
argparams[argparams['loss']])
del argparams[argparams['loss']]
del argparams['loss']
else:
loss_args = "--loss {}".format(argparams["loss"])
del argparams['loss']
if not utils.file_exists(data.file_edges):
data.to_edgelist()
with utils.TempDir() as tmp_dir:
pajek = utils.link_file(data.file_edges,
dest_dir=tmp_dir,
destname='edges.txt')
cmdargs = ["--{} {}".format(u, v) for u, v in argparams.items()]
cmdargs.append(loss_args)
cmdargs.append("-o lsooutput")
cmdargs = " ".join(cmdargs)
cmd = "{} {} {}".format(config.LSO_CLUSTER_PROG, pajek, cmdargs)
with open(os.path.join(tmp_dir, "tmpcmd"), 'wt') as f:
f.write(cmd)
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait("bash -x ./tmpcmd", tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
outputfiles = tmp_dir + "/lsooutput"
import pandas as pd
output = pd.read_csv(outputfiles, sep="\t", header=None)
output.columns = ['node', 'cluster']
clusters = output
clusters = clusters.groupby('cluster').apply(
lambda u: list(u['node'])).to_dict()
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self,
data,
seed=None,
thread=None,
threshold=1e-3,
vThresh=0.0,
maxIter=10,
prior=True):
if seed is None:
seed = np.random.randint(999999)
params = locals()
del params['self']
del params['data']
params = {u: v for u, v in params.items() if v is not None}
thread = utils.get_num_thread(thread)
params['thread'] = thread
if (data.is_directed() or data.is_weighted()) and False:
raise Exception(
"only undirected and unweighted graph is supported")
if not utils.file_exists(data.file_edges):
data.to_edgelist()
with utils.TempDir() as tmp_dir:
pajek = utils.link_file(data.file_edges, tmp_dir, 'edges.txt')
cmd = "{prog} {seed} {graph} {thread} 1 {threshold} {vThresh} {maxIter} {output_dir} {prior}"\
.format(prog=config.get_powergraph_prog('RelaxMap', data.is_directed()), seed=seed,
graph=pajek, thread=thread, threshold=threshold, vThresh=vThresh, output_dir=tmp_dir,
maxIter=maxIter, prior='prior' if prior else 'normal').strip()
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
outputfile = os.path.join(tmp_dir, "edges.tree")
import pandas as pd
output = pd.read_csv(outputfile, sep=" ", header=None, skiprows=1)
output['node'] = output.loc[:, 2].astype(np.int)
output['cluster'] = output.loc[:, 0].map(
lambda u: u.split(':')[0]).astype(np.int)
clusters = output[['cluster', 'node']]
clusters = clusters.groupby('cluster').apply(
lambda u: list(u['node'])).to_dict()
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
