Source code for pypath.internals.refs

#!/usr/bin/env python
# -*- coding: utf-8 -*-

#
#  This file is part of the `pypath` python module
#
#  Copyright 2014-2023
#  EMBL, EMBL-EBI, Uniklinik RWTH Aachen, Heidelberg University
#
#  Authors: see the file `README.rst`
#  Contact: Dénes Türei (turei.denes@gmail.com)
#
#  Distributed under the GPLv3 License.
#  See accompanying file LICENSE.txt or copy at
#      https://www.gnu.org/licenses/gpl-3.0.html
#
#  Website: https://pypath.omnipathdb.org/
#

from future.utils import iteritems
import os
import sys
import webbrowser
import pandas as pd
try:
    import cPickle as pickle
except:
    import pickle

import pypath.share.curl as curl
import pypath.share.common as common
import pypath.resources.urls as urls
from pypath.inputs import pubmed as pubmed_input
import pypath.share.cache as cache
import pypath.inputs.pubmed as pubmed


[docs] class Reference(object): __slots__ = ['pmid']
[docs] def __init__(self, pmid): self.pmid = str(pmid).strip()
def __eq__(self, other): return self.pmid == other.pmid def __hash__(self): return hash(self.pmid) def open(self): pubmed_input.open_pubmed(self.pmid) def __str__(self): return self.pmid def info(self): return pubmed_input.get_pubmeds([self.pmid]) def __repr__(self): return '<Reference: %s>' % self.pmid
[docs] 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 = cache.cache_item('pubmed_cache') if htp_threshold is not None: pp.htp_stats() pubmeds = common.unique_list( common.flat_list([[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 = pubmed_input.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.unique_list(points) earliest = common.unique_list(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