kkonganti@0: """Reports and visualizes results"""
kkonganti@0: 
kkonganti@0: import logging, os, pandas, re, shutil, time
kkonganti@0: import matplotlib.pyplot as plt
kkonganti@0: import seaborn as sns
kkonganti@0: import lexmapr.ontology_reasoner as ontr
kkonganti@0: 
kkonganti@0: logging.getLogger('matplotlib').setLevel(logging.WARNING)
kkonganti@0: 
kkonganti@0: 
kkonganti@0: def _split_results(pandas_series, x_col, y_col, split_delim=True):
kkonganti@0:     '''Format a value count series to a dataframe, spliting |-delimited terms'''
kkonganti@0:     graph_dic = {}
kkonganti@0:     for x in pandas_series.items():
kkonganti@0:         for y in x[0].split('|'):
kkonganti@0:             try:
kkonganti@0:                 graph_dic[y] += x[1]
kkonganti@0:             except(KeyError):
kkonganti@0:                 graph_dic[y] = x[1]
kkonganti@0:     if split_delim:
kkonganti@0:         graph_pd=pandas.DataFrame({x_col:[':'.join(x.split(':')[:-1]) for x in graph_dic.keys()],
kkonganti@0:                                    y_col:list(graph_dic.values())})
kkonganti@0:     else:
kkonganti@0:         graph_pd=pandas.DataFrame({x_col:list(graph_dic.keys()),
kkonganti@0:                                    y_col:list(graph_dic.values())})
kkonganti@0:     return(graph_pd)
kkonganti@0: 
kkonganti@0: 
kkonganti@0: def _get_ontols(map_res, match_col, bin_col):
kkonganti@0:     '''Make instances of Ontology_accessions and group as relevant'''
kkonganti@0:     red_res = map_res[map_res[bin_col].notna()]
kkonganti@0:     mapped_terms = _split_results(red_res[match_col].value_counts(), 'x', 'y', split_delim=False)
kkonganti@0:     mapped_bins = _split_results(red_res[bin_col].value_counts(), 'x', 'y', split_delim=False)
kkonganti@0:     ontol_sets = {}
kkonganti@0:     lcp_set = set()
kkonganti@0:     term_set = set()
kkonganti@0:     for y in list(mapped_bins['x']):
kkonganti@0:         ontol_sets[ontr.Ontology_accession.make_instance(y)] = set()
kkonganti@0:         time.sleep(0.05)
kkonganti@0:     for x in list(mapped_terms['x']):
kkonganti@0:         if x == 'No Match':
kkonganti@0:             continue
kkonganti@0:         term_ontol = ontr.Ontology_accession.make_instance(x)
kkonganti@0:         if term_ontol.ancestors == 'not assigned yet':
kkonganti@0:             term_ontol.get_family('ancestors')
kkonganti@0:             time.sleep(0.05)
kkonganti@0:         if term_ontol.ancestors == ['none found']:
kkonganti@0:             continue
kkonganti@0:         for y in ontol_sets:
kkonganti@0:             if y in term_ontol.ancestors:
kkonganti@0:                 ontol_sets[y].add(term_ontol)
kkonganti@0:     for y in ontol_sets:
kkonganti@0:        if ontol_sets[y] != set():
kkonganti@0:           lcp_set.add(y)
kkonganti@0:           term_set = term_set | ontol_sets[y]
kkonganti@0:     if len(term_set) > 100:
kkonganti@0:         term_list = [x.id for x in list(term_set)]
kkonganti@0:         terms_string = ''
kkonganti@0:         for a,b,c,d in zip(term_list[::4],term_list[1::4],term_list[2::4],term_list[3::4]):
kkonganti@0:             terms_string += f'\n\t\t{a}\t{b}\t{c}\t{d}'
kkonganti@0:         logging.info(f'Not drawing {bin_col} graph with {len(term_list)} child nodes:\n\
kkonganti@0:                        {terms_string}\n')
kkonganti@0:         return([],[])
kkonganti@0:     return(list(lcp_set), list(term_set))
kkonganti@0: 
kkonganti@0: 
kkonganti@0: def report_results(out_file, arg_bins):
kkonganti@0:     '''Print mapping counts to log'''
kkonganti@0:     mapping_results = pandas.read_csv(out_file, header=0, delimiter='\t')
kkonganti@0:     match_status = mapping_results['Match_Status (Macro Level)'].value_counts()
kkonganti@0:     logging.info(f'\t\tNo. unique terms: '+str(len(mapping_results['Sample_Desc'])))
kkonganti@0:     for x in match_status.items():
kkonganti@0:         logging.info(f'\t\tNo. {x[0]}: {x[1]}')
kkonganti@0:     for x in arg_bins:
kkonganti@0:         logging.info(f'\t\tNo. mapped under {x}: {mapping_results[x].count()}')
kkonganti@0: 
kkonganti@0: 
kkonganti@0: def report_cache(term_cache):
kkonganti@0:     # TODO: add counts for bins?
kkonganti@0:     '''Print mapping counts to log from cache, only count unique terms'''
kkonganti@0:     logging.info(f'\t\tNo. unique terms: {len(term_cache)-1}')
kkonganti@0:     no_match = 0
kkonganti@0:     full_match = 0
kkonganti@0:     syno_match = 0
kkonganti@0:     comp_match = 0
kkonganti@0:     for x in term_cache:
kkonganti@0:         if re.search('No Match', term_cache[x]):
kkonganti@0:             no_match += 1
kkonganti@0:         if re.search('Full Term Match', term_cache[x]):
kkonganti@0:             full_match += 1
kkonganti@0:         if re.search('Synonym Match', term_cache[x]):
kkonganti@0:             syno_match += 1
kkonganti@0:         if re.search('Component Match', term_cache[x]):
kkonganti@0:             comp_match += 1
kkonganti@0:     logging.info(f'\t\tNo. Unique Full Term Match: {full_match}')
kkonganti@0:     logging.info(f'\t\tNo. Unique Synonym Match: {syno_match}')
kkonganti@0:     logging.info(f'\t\tNo. Unique Component Match: {comp_match}')
kkonganti@0:     logging.info(f'\t\tNo. Unique No Match: {no_match}')
kkonganti@0:     return({'No Match':no_match, 'Full Term Match':full_match,
kkonganti@0:             'Synonym Match':syno_match, 'Component Match':comp_match})
kkonganti@0: 
kkonganti@0: 
kkonganti@0: def figure_folder():
kkonganti@0:     '''Prepare figures folder'''
kkonganti@0:     try:
kkonganti@0:         shutil.rmtree('lexmapr_figures/')
kkonganti@0:     except(FileNotFoundError):
kkonganti@0:         pass
kkonganti@0:     os.mkdir('lexmapr_figures/')
kkonganti@0: 
kkonganti@0: 
kkonganti@0: def visualize_cache(match_counts):
kkonganti@0:     '''Generate graph'''
kkonganti@0:     # TODO: add graphing for bins?
kkonganti@0:     x_col = 'Match status'
kkonganti@0:     y_col = 'No. samples matched'
kkonganti@0:     sns_fig = sns.barplot(x=list(match_counts.keys()),
kkonganti@0:                           y=list(match_counts.values()), ci=None).get_figure()
kkonganti@0:     plt.xticks(rotation=90)
kkonganti@0:     plt.tight_layout()
kkonganti@0:     sns_fig.savefig('lexmapr_figures/mapping_results.png')
kkonganti@0:     logging.info(f'Did not attempt to make bin graphs')
kkonganti@0: 
kkonganti@0: 
kkonganti@0: def visualize_results(out_file, arg_bins):
kkonganti@0:     '''Generate graphs'''
kkonganti@0:     map_res = pandas.read_csv(out_file,delimiter='\t')
kkonganti@0:     x_col = 'Match status'
kkonganti@0:     y_col = 'No. samples matched'
kkonganti@0:     match_status = map_res['Match_Status (Macro Level)'].value_counts()
kkonganti@0:     match_res = _split_results(match_status, x_col, y_col, False)
kkonganti@0:     match_res = match_res.sort_values(y_col,ascending=False)
kkonganti@0:     sns_fig = sns.barplot(x=x_col, y=y_col, data=match_res, ci=None).get_figure()
kkonganti@0:     plt.xticks(rotation=90)
kkonganti@0:     plt.tight_layout()
kkonganti@0:     sns_fig.savefig('lexmapr_figures/mapping_results.png')
kkonganti@0: 
kkonganti@0:     if map_res.shape[0] >= 1000:
kkonganti@0:         logging.info(f'Did not attempt to make bin because too many rows')
kkonganti@0:         return
kkonganti@0: 
kkonganti@0:     if arg_bins != []:
kkonganti@0:         x_col = 'Bin'
kkonganti@0:         bin_counts = {}
kkonganti@0:         for x in arg_bins:
kkonganti@0:             bin_counts[x] = sum(map_res[x].value_counts())
kkonganti@0:             bin_res = _split_results(map_res[x].value_counts(), x_col, y_col)
kkonganti@0:             if not bin_res.empty:
kkonganti@0:                 bin_res = bin_res.sort_values(y_col,ascending=False)
kkonganti@0:                 plt.clf()
kkonganti@0:                 sns_fig = sns.barplot(x=x_col, y=y_col, data=bin_res, ci=None).get_figure()
kkonganti@0:                 plt.xticks(rotation=90)
kkonganti@0:                 plt.tight_layout()
kkonganti@0:                 plt.savefig(f'lexmapr_figures/{x}_binning.png')
kkonganti@0: 
kkonganti@0:         plt.clf()
kkonganti@0:         bin_pd = pandas.DataFrame({x_col:list(bin_counts.keys()), 
kkonganti@0:                                    y_col:list(bin_counts.values())})
kkonganti@0:         bin_pd = bin_pd.sort_values(y_col,ascending=False)
kkonganti@0:         sns_fig = sns.barplot(x=x_col, y=y_col, data=bin_pd, ci=None).get_figure()
kkonganti@0:         plt.xticks(rotation=90)
kkonganti@0:         plt.tight_layout()
kkonganti@0:         sns_fig.savefig('lexmapr_figures/binning_results.png')
kkonganti@0:         
kkonganti@0:         # TODO: make node colors vary with frequency and color ones that are both top and bottom?
kkonganti@0:         for x in arg_bins:
kkonganti@0:             print(f'\tMight generate {x} ontology graph...'.ljust(80),end='\r')
kkonganti@0:             lcp_list, term_list = _get_ontols(map_res, 'Matched_Components', x)
kkonganti@0:             if lcp_list != [] and term_list != []:
kkonganti@0:                 bin_package = ontr.Ontology_package('.', list(term_list))
kkonganti@0:                 bin_package.set_lcp(lcp_list)
kkonganti@0:                 bin_package.visualize_terms(f'lexmapr_figures/{x}_terms.png',
kkonganti@0:                                             show_lcp=True, fill_out=True, trim_nodes=True)