def interactions(self, req):
fields = [b'sources', b'references']
result = []
elist = self._get_eids(req)
res = []
hdr = [
'source', 'target', 'is_directed', 'is_stimulation',
'is_inhibition'
]
if b'fields' in req.args:
hdr += [
f.decode('utf-8') for f in fields if f in req.args[b'fields']
]
all_sources = set([])
for eid in elist:
e = self.g.es[eid]
all_sources = all_sources | e['sources']
for d in ['straight', 'reverse']:
uniprots = getattr(e['dirs'], d)
if e['dirs'].dirs[uniprots]:
thisEdge = [
uniprots[0], uniprots[1], 1,
int(e['dirs'].is_stimulation(uniprots)),
int(e['dirs'].is_inhibition(uniprots))
]
dsources = e['dirs'].get_dir(uniprots, sources=True)
dsources = dsources | e['dirs'].get_dir('undirected',
sources=True)
if 'sources' in hdr:
thisEdge.append(list(dsources))
if 'references' in hdr:
thisEdge.append([
r.pmid for r in flatList([
rs for s, rs in iteritems(e['refs_by_source'])
if s in dsources
])
])
thisEdge.append(self._dip_urls(e))
res.append(thisEdge)
if not e['dirs'].is_directed():
thisEdge = [e['dirs'].nodes[0], e['dirs'].nodes[1], 0, 0, 0]
if 'sources' in hdr:
thisEdge.append(list(e['sources']))
if 'references' in hdr:
thisEdge.append([r.pmid for r in e['references']])
thisEdge.append(self._dip_urls(e))
res.append(thisEdge)
if 'DIP' in all_sources:
hdr.append('dip_url')
else:
res = map(lambda r: r[:-1], res)
if b'format' in req.args and req.args[b'format'] == b'json':
return json.dumps([dict(zip(hdr, r)) for r in res])
else:
return self._table_output(res, hdr, req)
def toplist(self,
length=None,
alpha=None,
significant=True,
min_set_size=0,
groups=None,
filtr=lambda x: True,
**kwargs):
args = get_args(locals(), ['filtr', 'groups'])
if groups is None:
groups = self.gsea.groups.keys() # all by default
sets = set(
common.flatList(s for g, s in iteritems(self.gsea.groups)
if g in groups))
return super(GSEABinaryEnrichmentSet,
self).toplist(filtr=lambda x: x[0] in sets and filtr(x),
**args)
def write_set(self, id_list, setname, id_type, map_ids=True):
self.sets[setname] = set(common.uniqList(common.flatList(
self.mapper.map_name(n, self.ids[id_type], self.target_id)
for n in id_list))) if map_ids \
else set(id_list)
def get_pubmed_data(pp, cachefile=None, htp_threshold=20):
"""
For one PyPath object, obtains metadata for all PubMed IDs
through NCBI E-utils.
:param pp:
``pypath.PyPath`` object
:param htp_threshold:
The number of interactions for one reference
above the study considered to be high-throughput.
"""
if cachefile is None:
cachefile = settings.get('pubmed_cache')
if htp_threshold is not None:
pp.htp_stats()
pubmeds = common.uniqList(
common.flatList([[r.pmid for r in e['references']]
for e in pp.graph.es]))
if htp_threshold is not None:
pubmeds = set(pubmeds) - pp.htp[htp_threshold]['htrefs']
notpmid = [i for i in pubmeds if not i.isdigit()]
sys.stdout.write('\t:: Number of non PubMed ID references: %u\n' %
len(notpmid))
pmdata = {}
if os.path.exists(cachefile):
sys.stdout.write('\t:: Loading data previously downloaded '
'from PubMed, from file `%s`\n' % cachefile)
pmdata = pickle.load(open(cachefile, 'rb'))
missing = list(set(pubmeds) - set(pmdata.keys()))
sys.stdout.write('\t:: Downloading data from PubMed about %s papers\n' %
len(missing))
cached_pubmeds_len = len(pmdata)
pmdata_new = dataio.get_pubmeds(missing)
pmdata.update(pmdata_new)
sys.stdout.write('\t:: Saving PubMed data to file `%s`\n' % cachefile)
if len(pmdata) > cached_pubmeds_len:
pickle.dump(pmdata, open(cachefile, 'wb'))
pmdata = dict(i for i in pmdata.items() if i[0] in pubmeds)
points = []
earliest = []
for e in pp.graph.es:
for s, rs in iteritems(e['refs_by_source']):
pms = [
r.pmid for r in rs
if (htp_threshold is None
or r.pmid not in pp.htp[htp_threshold]['htrefs'])
and r.pmid in pmdata and 'pubdate' in pmdata[r.pmid]
]
if len(pms) > 0:
yrs = [int(pmdata[pm]['pubdate'][:4]) for pm in pms]
earliest.append((s, 0, min(yrs), '', e.index))
for pm in pms:
points.append((s, pm, int(pmdata[pm]['pubdate'][:4]),
pmdata[pm]['source'], e.index))
points = common.uniqList(points)
earliest = common.uniqList(earliest)
points = pd.DataFrame.from_records(points)
earliest = pd.DataFrame.from_records(earliest)
points.columns = ['database', 'pmid', 'year', 'journal', 'eid']
earliest.columns = ['database', 'none', 'year', 'none', 'eid']
return points, earliest
