def main(argv=None):
"""script main.
parses command line options in sys.argv, unless *argv* is given.
"""
if argv is None:
argv = sys.argv
# setup command line parser
parser = E.OptionParser(version="%prog version: $Id$",
usage=globals()["__doc__"])
parser.add_option("-t",
"--test",
dest="test",
type="string",
help="supply help")
parser.add_option("--method",
dest="method",
type="choice",
choices=("metrics", "summary", "module_summary"),
help="method to summarise clustering")
parser.add_option("--ref-gtf-files",
dest="ref_gtf",
type="string",
help="comma separated list of reference gtf files")
# add common options (-h/--help, ...) and parse command line
(options, args) = E.Start(parser, argv=argv)
if options.method == "metrics":
infile = argv[-1]
E.info("loading input file: %s" % infile)
assert infile
df = pd.read_table(infile, sep="\t", header=None, index_col=0)
df = df.ix[:, :50]
cluster_combs = (x for x in itertools.combinations(df.columns, 2))
genes = df.index
results_dict = {}
all_clusts = {}
E.info("setting up cluster containers")
for i in df.columns:
clusters = set(df[i].values.tolist())
cluster_dict = {}
for clust in clusters:
cluster_dict[clust] = []
for gene in genes:
cluster_dict[df[i][gene]].append(gene)
for col in clusters:
col_set = set()
clust_col = cluster_dict[col]
gene_members = itertools.combinations(clust_col, 2)
col_set.update(gene_members)
cluster_dict[col] = col_set
all_clusts[i] = cluster_dict
E.info("generating all pair-wise cluster comparisons")
E.info("calculating adjusted mutual information")
for k in cluster_combs:
clusters1 = all_clusts[k[0]]
clusters2 = all_clusts[k[1]]
metric_dict = {}
metric_dict['AMI'] = TS.adjustedMutualInformation(
clusters1, clusters2)
results_dict[k] = metric_dict
res_frame = pd.DataFrame(results_dict).T
res_frame = res_frame.reset_index()
res_frame.drop(['level_0'], inplace=True, axis=1)
res_frame.drop(['level_1'], inplace=True, axis=1)
# flatten rand indices and add to output dataframe
rand_arrays = TS.randIndexes(df)
flat_adj_rand = TS.unravel_arrays(rand_arrays[0])
flat_rand = TS.unravel_arrays(rand_arrays[1])
res_frame['Rand_Index'] = flat_rand
res_frame['Adjusted_Rand_Index'] = flat_adj_rand
E.info("aggregating results")
res_frame.to_csv(options.stdout, sep="\t", index_label='idx')
elif options.method == "summary":
infiles = argv[-1]
list_of_files = infiles.split(",")
file_dict = {}
for fle in list_of_files:
fname = fle.split("/")[-1]
condition = fname.split("-")[0]
ref = fname.split("-")[1]
df_ = pd.read_table(fle, sep="\t", header=0, index_col=0)
df_.columns = ['gene_id', 'cluster']
clust_dict = {}
for idx in df_.index:
cluster = df_.loc[idx]['cluster']
gene = df_.loc[idx]['gene_id']
try:
clust_dict[cluster] += 1
except KeyError:
clust_dict[cluster] = 1
med_size = np.median(list(clust_dict.values()))
file_dict[fname] = {
'condition': condition,
'reference': ref,
'median_cluster_size': med_size
}
outframe = pd.DataFrame(file_dict).T
outframe.to_csv(options.stdout, sep="\t", index_label='idx')
elif options.method == "module_summary":
# get lncRNA/gene lengths from reference gtfs
ref_gtfs = options.ref_gtf.split(",")
length_dict = {}
for ref in ref_gtfs:
oref = IOTools.openFile(ref, "rb")
git = GTF.transcript_iterator(GTF.iterator(oref))
for gene in git:
for trans in gene:
length = trans.end - trans.start
try:
length_dict[trans.gene_id] += length
except KeyError:
length_dict[trans.gene_id] = length
oref.close()
infiles = argv[-1]
list_of_files = infiles.split(",")
fdfs = []
for fle in list_of_files:
cond = fle.split("/")[-1].split("-")[0]
refer = fle.split("/")[-1].split("-")[1]
_df = pd.read_table(fle, sep="\t", header=0, index_col=0)
_df.columns = ['gene_id', 'cluster']
clusters = set(_df['cluster'])
c_dict = {}
# summarize over each cluster
for clust in clusters:
lengths = []
c_df = _df[_df['cluster'] == clust]
for lid in c_df['gene_id']:
lengths.append(length_dict[lid])
c_dict[clust] = {
'cluster_size': len(c_df['gene_id']),
'mean_length': np.mean(lengths),
'index': (cond, refer),
'module': clust
}
cdf = pd.DataFrame(c_dict).T
# use a multindex for hierarchical indexing
midx = pd.MultiIndex.from_tuples(cdf['index'])
cdf.index = midx
cdf.drop(['index'], inplace=True, axis=1)
fdfs.append(cdf)
# generate a single output df
s_df = fdfs[0]
fdfs.pop(0)
for df in fdfs:
s_df = s_df.append(df)
s_df.to_csv(options.stdout,
index_label=("condition", "reference"),
sep="\t")
# write footer and output benchmark information.
E.Stop()
def main(argv=None):
"""script main.
parses command line options in sys.argv, unless *argv* is given.
"""
if argv is None:
argv = sys.argv
# setup command line parser
parser = E.OptionParser(version="%prog version: $Id$",
usage=globals()["__doc__"])
parser.add_option("-t", "--test", dest="test", type="string",
help="supply help")
parser.add_option("--method", dest="method", type="choice",
choices=("metrics", "summary", "module_summary"),
help="method to summarise clustering")
parser.add_option("--ref-gtf-files", dest="ref_gtf", type="string",
help="comma separated list of reference gtf files")
# add common options (-h/--help, ...) and parse command line
(options, args) = E.Start(parser, argv=argv)
if options.method == "metrics":
infile = argv[-1]
E.info("loading input file: %s" % infile)
assert infile
df = pd.read_table(infile,
sep="\t",
header=None,
index_col=0)
df = df.ix[:, :50]
cluster_combs = (x for x in itertools.combinations(df.columns,
2))
genes = df.index
results_dict = {}
all_clusts = {}
E.info("setting up cluster containers")
for i in df.columns:
clusters = set(df[i].values.tolist())
cluster_dict = {}
for clust in clusters:
cluster_dict[clust] = []
for gene in genes:
cluster_dict[df[i][gene]].append(gene)
for col in clusters:
col_set = set()
clust_col = cluster_dict[col]
gene_members = itertools.combinations(clust_col,
2)
col_set.update(gene_members)
cluster_dict[col] = col_set
all_clusts[i] = cluster_dict
E.info("generating all pair-wise cluster comparisons")
E.info("calculating adjusted mutual information")
for k in cluster_combs:
clusters1 = all_clusts[k[0]]
clusters2 = all_clusts[k[1]]
metric_dict = {}
metric_dict['AMI'] = TS.adjustedMutualInformation(clusters1,
clusters2)
results_dict[k] = metric_dict
res_frame = pd.DataFrame(results_dict).T
res_frame = res_frame.reset_index()
res_frame.drop(['level_0'], inplace=True, axis=1)
res_frame.drop(['level_1'], inplace=True, axis=1)
# flatten rand indices and add to output dataframe
rand_arrays = TS.randIndexes(df)
flat_adj_rand = TS.unravel_arrays(rand_arrays[0])
flat_rand = TS.unravel_arrays(rand_arrays[1])
res_frame['Rand_Index'] = flat_rand
res_frame['Adjusted_Rand_Index'] = flat_adj_rand
E.info("aggregating results")
res_frame.to_csv(options.stdout,
sep="\t",
index_label='idx')
elif options.method == "summary":
infiles = argv[-1]
list_of_files = infiles.split(",")
file_dict = {}
for fle in list_of_files:
fname = fle.split("/")[-1]
condition = fname.split("-")[0]
ref = fname.split("-")[1]
df_ = pd.read_table(fle,
sep="\t",
header=0,
index_col=0)
df_.columns = ['gene_id', 'cluster']
clust_dict = {}
for idx in df_.index:
cluster = df_.loc[idx]['cluster']
gene = df_.loc[idx]['gene_id']
try:
clust_dict[cluster] += 1
except KeyError:
clust_dict[cluster] = 1
med_size = np.median(clust_dict.values())
file_dict[fname] = {'condition': condition,
'reference': ref,
'median_cluster_size': med_size}
outframe = pd.DataFrame(file_dict).T
outframe.to_csv(options.stdout,
sep="\t",
index_label='idx')
elif options.method == "module_summary":
# get lncRNA/gene lengths from reference gtfs
ref_gtfs = options.ref_gtf.split(",")
length_dict = {}
for ref in ref_gtfs:
oref = IOTools.openFile(ref, "rb")
git = GTF.transcript_iterator(GTF.iterator(oref))
for gene in git:
for trans in gene:
length = trans.end - trans.start
try:
length_dict[trans.gene_id] += length
except KeyError:
length_dict[trans.gene_id] = length
oref.close()
infiles = argv[-1]
list_of_files = infiles.split(",")
fdfs = []
for fle in list_of_files:
cond = fle.split("/")[-1].split("-")[0]
refer = fle.split("/")[-1].split("-")[1]
_df = pd.read_table(fle, sep="\t",
header=0, index_col=0)
_df.columns = ['gene_id', 'cluster']
clusters = set(_df['cluster'])
c_dict = {}
# summarize over each cluster
for clust in clusters:
lengths = []
c_df = _df[_df['cluster'] == clust]
for lid in c_df['gene_id']:
lengths.append(length_dict[lid])
c_dict[clust] = {'cluster_size': len(c_df['gene_id']),
'mean_length': np.mean(lengths),
'index': (cond, refer),
'module': clust}
cdf = pd.DataFrame(c_dict).T
# use a multindex for hierarchical indexing
midx = pd.MultiIndex.from_tuples(cdf['index'])
cdf.index = midx
cdf.drop(['index'], inplace=True, axis=1)
fdfs.append(cdf)
# generate a single output df
s_df = fdfs[0]
fdfs.pop(0)
for df in fdfs:
s_df = s_df.append(df)
s_df.to_csv(options.stdout,
index_label=("condition", "reference"),
sep="\t")
# write footer and output benchmark information.
E.Stop()
