Mercurial > repos > kkonganti > cfsan_lexmapr2

comparison lexmapr/run_summary.py @ 0:f5c39d0447be

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload"
author kkonganti
date Wed, 31 Aug 2022 14:32:07 -0400
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:f5c39d0447be
1 """Reports and visualizes results"""
2
3 import logging, os, pandas, re, shutil, time
4 import matplotlib.pyplot as plt
5 import seaborn as sns
6 import lexmapr.ontology_reasoner as ontr
7
8 logging.getLogger('matplotlib').setLevel(logging.WARNING)
9
10
11 def _split_results(pandas_series, x_col, y_col, split_delim=True):
12 '''Format a value count series to a dataframe, spliting |-delimited terms'''
13 graph_dic = {}
14 for x in pandas_series.items():
15 for y in x[0].split('|'):
16 try:
17 graph_dic[y] += x[1]
18 except(KeyError):
19 graph_dic[y] = x[1]
20 if split_delim:
21 graph_pd=pandas.DataFrame({x_col:[':'.join(x.split(':')[:-1]) for x in graph_dic.keys()],
22 y_col:list(graph_dic.values())})
23 else:
24 graph_pd=pandas.DataFrame({x_col:list(graph_dic.keys()),
25 y_col:list(graph_dic.values())})
26 return(graph_pd)
27
28
29 def _get_ontols(map_res, match_col, bin_col):
30 '''Make instances of Ontology_accessions and group as relevant'''
31 red_res = map_res[map_res[bin_col].notna()]
32 mapped_terms = _split_results(red_res[match_col].value_counts(), 'x', 'y', split_delim=False)
33 mapped_bins = _split_results(red_res[bin_col].value_counts(), 'x', 'y', split_delim=False)
34 ontol_sets = {}
35 lcp_set = set()
36 term_set = set()
37 for y in list(mapped_bins['x']):
38 ontol_sets[ontr.Ontology_accession.make_instance(y)] = set()
39 time.sleep(0.05)
40 for x in list(mapped_terms['x']):
41 if x == 'No Match':
42 continue
43 term_ontol = ontr.Ontology_accession.make_instance(x)
44 if term_ontol.ancestors == 'not assigned yet':
45 term_ontol.get_family('ancestors')
46 time.sleep(0.05)
47 if term_ontol.ancestors == ['none found']:
48 continue
49 for y in ontol_sets:
50 if y in term_ontol.ancestors:
51 ontol_sets[y].add(term_ontol)
52 for y in ontol_sets:
53 if ontol_sets[y] != set():
54 lcp_set.add(y)
55 term_set = term_set | ontol_sets[y]
56 if len(term_set) > 100:
57 term_list = [x.id for x in list(term_set)]
58 terms_string = ''
59 for a,b,c,d in zip(term_list[::4],term_list[1::4],term_list[2::4],term_list[3::4]):
60 terms_string += f'\n\t\t{a}\t{b}\t{c}\t{d}'
61 logging.info(f'Not drawing {bin_col} graph with {len(term_list)} child nodes:\n\
62 {terms_string}\n')
63 return([],[])
64 return(list(lcp_set), list(term_set))
65
66
67 def report_results(out_file, arg_bins):
68 '''Print mapping counts to log'''
69 mapping_results = pandas.read_csv(out_file, header=0, delimiter='\t')
70 match_status = mapping_results['Match_Status (Macro Level)'].value_counts()
71 logging.info(f'\t\tNo. unique terms: '+str(len(mapping_results['Sample_Desc'])))
72 for x in match_status.items():
73 logging.info(f'\t\tNo. {x[0]}: {x[1]}')
74 for x in arg_bins:
75 logging.info(f'\t\tNo. mapped under {x}: {mapping_results[x].count()}')
76
77
78 def report_cache(term_cache):
79 # TODO: add counts for bins?
80 '''Print mapping counts to log from cache, only count unique terms'''
81 logging.info(f'\t\tNo. unique terms: {len(term_cache)-1}')
82 no_match = 0
83 full_match = 0
84 syno_match = 0
85 comp_match = 0
86 for x in term_cache:
87 if re.search('No Match', term_cache[x]):
88 no_match += 1
89 if re.search('Full Term Match', term_cache[x]):
90 full_match += 1
91 if re.search('Synonym Match', term_cache[x]):
92 syno_match += 1
93 if re.search('Component Match', term_cache[x]):
94 comp_match += 1
95 logging.info(f'\t\tNo. Unique Full Term Match: {full_match}')
96 logging.info(f'\t\tNo. Unique Synonym Match: {syno_match}')
97 logging.info(f'\t\tNo. Unique Component Match: {comp_match}')
98 logging.info(f'\t\tNo. Unique No Match: {no_match}')
99 return({'No Match':no_match, 'Full Term Match':full_match,
100 'Synonym Match':syno_match, 'Component Match':comp_match})
101
102
103 def figure_folder():
104 '''Prepare figures folder'''
105 try:
106 shutil.rmtree('lexmapr_figures/')
107 except(FileNotFoundError):
108 pass
109 os.mkdir('lexmapr_figures/')
110
111
112 def visualize_cache(match_counts):
113 '''Generate graph'''
114 # TODO: add graphing for bins?
115 x_col = 'Match status'
116 y_col = 'No. samples matched'
117 sns_fig = sns.barplot(x=list(match_counts.keys()),
118 y=list(match_counts.values()), ci=None).get_figure()
119 plt.xticks(rotation=90)
120 plt.tight_layout()
121 sns_fig.savefig('lexmapr_figures/mapping_results.png')
122 logging.info(f'Did not attempt to make bin graphs')
123
124
125 def visualize_results(out_file, arg_bins):
126 '''Generate graphs'''
127 map_res = pandas.read_csv(out_file,delimiter='\t')
128 x_col = 'Match status'
129 y_col = 'No. samples matched'
130 match_status = map_res['Match_Status (Macro Level)'].value_counts()
131 match_res = _split_results(match_status, x_col, y_col, False)
132 match_res = match_res.sort_values(y_col,ascending=False)
133 sns_fig = sns.barplot(x=x_col, y=y_col, data=match_res, ci=None).get_figure()
134 plt.xticks(rotation=90)
135 plt.tight_layout()
136 sns_fig.savefig('lexmapr_figures/mapping_results.png')
137
138 if map_res.shape[0] >= 1000:
139 logging.info(f'Did not attempt to make bin because too many rows')
140 return
141
142 if arg_bins != []:
143 x_col = 'Bin'
144 bin_counts = {}
145 for x in arg_bins:
146 bin_counts[x] = sum(map_res[x].value_counts())
147 bin_res = _split_results(map_res[x].value_counts(), x_col, y_col)
148 if not bin_res.empty:
149 bin_res = bin_res.sort_values(y_col,ascending=False)
150 plt.clf()
151 sns_fig = sns.barplot(x=x_col, y=y_col, data=bin_res, ci=None).get_figure()
152 plt.xticks(rotation=90)
153 plt.tight_layout()
154 plt.savefig(f'lexmapr_figures/{x}_binning.png')
155
156 plt.clf()
157 bin_pd = pandas.DataFrame({x_col:list(bin_counts.keys()),
158 y_col:list(bin_counts.values())})
159 bin_pd = bin_pd.sort_values(y_col,ascending=False)
160 sns_fig = sns.barplot(x=x_col, y=y_col, data=bin_pd, ci=None).get_figure()
161 plt.xticks(rotation=90)
162 plt.tight_layout()
163 sns_fig.savefig('lexmapr_figures/binning_results.png')
164
165 # TODO: make node colors vary with frequency and color ones that are both top and bottom?
166 for x in arg_bins:
167 print(f'\tMight generate {x} ontology graph...'.ljust(80),end='\r')
168 lcp_list, term_list = _get_ontols(map_res, 'Matched_Components', x)
169 if lcp_list != [] and term_list != []:
170 bin_package = ontr.Ontology_package('.', list(term_list))
171 bin_package.set_lcp(lcp_list)
172 bin_package.visualize_terms(f'lexmapr_figures/{x}_terms.png',
173 show_lcp=True, fill_out=True, trim_nodes=True)