ligdis commited on
Commit
223c626
·
verified ·
1 Parent(s): 0f91c49

Upload 2 files

Browse files
Files changed (2) hide show
  1. fid.txt +400 -0
  2. proteome_meta.py +104 -0
fid.txt ADDED
@@ -0,0 +1,400 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ C001
2
+ C002
3
+ C003
4
+ C004
5
+ C006
6
+ C007
7
+ C008
8
+ C009
9
+ C010
10
+ C011
11
+ C012
12
+ C013
13
+ C014
14
+ C015
15
+ C017
16
+ C018
17
+ C020
18
+ C021
19
+ C022
20
+ C023
21
+ C024
22
+ C025
23
+ C026
24
+ C027
25
+ C028
26
+ C029
27
+ C030
28
+ C031
29
+ C032
30
+ C033
31
+ C034
32
+ C036
33
+ C037
34
+ C038
35
+ C039
36
+ C040
37
+ C041
38
+ C042
39
+ C044
40
+ C045
41
+ C046
42
+ C048
43
+ C049
44
+ C050
45
+ C051
46
+ C052
47
+ C053
48
+ C054
49
+ C055
50
+ C056
51
+ C057
52
+ C058
53
+ C059
54
+ C060
55
+ C061
56
+ C062
57
+ C063
58
+ C064
59
+ C065
60
+ C066
61
+ C067
62
+ C068
63
+ C070
64
+ C071
65
+ C072
66
+ C073
67
+ C074
68
+ C075
69
+ C076
70
+ C077
71
+ C078
72
+ C079
73
+ C080
74
+ C081
75
+ C082
76
+ C083
77
+ C084
78
+ C085
79
+ C086
80
+ C087
81
+ C088
82
+ C089
83
+ C091
84
+ C092
85
+ C093
86
+ C094
87
+ C095
88
+ C096
89
+ C097
90
+ C098
91
+ C099
92
+ C100
93
+ C101
94
+ C103
95
+ C104
96
+ C105
97
+ C106
98
+ C107
99
+ C108
100
+ C109
101
+ C110
102
+ C111
103
+ C112
104
+ C113
105
+ C114
106
+ C115
107
+ C116
108
+ C117
109
+ C119
110
+ C120
111
+ C121
112
+ C122
113
+ C123
114
+ C125
115
+ C126
116
+ C127
117
+ C128
118
+ C129
119
+ C130
120
+ C131
121
+ C132
122
+ C133
123
+ C134
124
+ C135
125
+ C137
126
+ C138
127
+ C139
128
+ C140
129
+ C141
130
+ C142
131
+ C143
132
+ C144
133
+ C145
134
+ C146
135
+ C147
136
+ C148
137
+ C149
138
+ C150
139
+ C151
140
+ C152
141
+ C153
142
+ C154
143
+ C156
144
+ C157
145
+ C158
146
+ C159
147
+ C160
148
+ C161
149
+ C162
150
+ C163
151
+ C164
152
+ C165
153
+ C166
154
+ C167
155
+ C168
156
+ C170
157
+ C171
158
+ C172
159
+ C173
160
+ C174
161
+ C176
162
+ C177
163
+ C178
164
+ C179
165
+ C180
166
+ C182
167
+ C183
168
+ C184
169
+ C185
170
+ C186
171
+ C187
172
+ C189
173
+ C190
174
+ C191
175
+ C193
176
+ C194
177
+ C195
178
+ C196
179
+ C197
180
+ C198
181
+ C199
182
+ C200
183
+ C201
184
+ C203
185
+ C204
186
+ C205
187
+ C207
188
+ C208
189
+ C210
190
+ C211
191
+ C212
192
+ C213
193
+ C214
194
+ C215
195
+ C216
196
+ C217
197
+ C218
198
+ C219
199
+ C220
200
+ C221
201
+ C222
202
+ C223
203
+ C225
204
+ C226
205
+ C227
206
+ C228
207
+ C229
208
+ C230
209
+ C231
210
+ C232
211
+ C233
212
+ C234
213
+ C235
214
+ C236
215
+ C237
216
+ C238
217
+ C239
218
+ C241
219
+ C242
220
+ C243
221
+ C244
222
+ C245
223
+ C246
224
+ C247
225
+ C248
226
+ C249
227
+ C250
228
+ C251
229
+ C252
230
+ C253
231
+ C254
232
+ C255
233
+ C256
234
+ C257
235
+ C258
236
+ C259
237
+ C260
238
+ C261
239
+ C262
240
+ C264
241
+ C265
242
+ C266
243
+ C268
244
+ C269
245
+ C270
246
+ C271
247
+ C272
248
+ C273
249
+ C275
250
+ C276
251
+ C277
252
+ C278
253
+ C279
254
+ C280
255
+ C281
256
+ C282
257
+ C283
258
+ C284
259
+ C285
260
+ C286
261
+ C287
262
+ C288
263
+ C289
264
+ C290
265
+ C291
266
+ C292
267
+ C293
268
+ C294
269
+ C295
270
+ C296
271
+ C297
272
+ C298
273
+ C299
274
+ C300
275
+ C301
276
+ C302
277
+ C303
278
+ C304
279
+ C305
280
+ C307
281
+ C308
282
+ C309
283
+ C310
284
+ C311
285
+ C312
286
+ C313
287
+ C314
288
+ C316
289
+ C317
290
+ C318
291
+ C319
292
+ C320
293
+ C321
294
+ C322
295
+ C323
296
+ C324
297
+ C326
298
+ C327
299
+ C328
300
+ C329
301
+ C330
302
+ C331
303
+ C332
304
+ C333
305
+ C334
306
+ C335
307
+ C336
308
+ C337
309
+ C338
310
+ C339
311
+ C341
312
+ C342
313
+ C343
314
+ C344
315
+ C345
316
+ C346
317
+ C347
318
+ C348
319
+ C349
320
+ C350
321
+ C351
322
+ C352
323
+ C353
324
+ C354
325
+ C355
326
+ C356
327
+ C357
328
+ C358
329
+ C359
330
+ C360
331
+ C361
332
+ C362
333
+ C363
334
+ C364
335
+ C365
336
+ C366
337
+ C367
338
+ C368
339
+ C369
340
+ C370
341
+ C371
342
+ C373
343
+ C374
344
+ C375
345
+ C376
346
+ C378
347
+ C379
348
+ C380
349
+ C381
350
+ C382
351
+ C383
352
+ C384
353
+ C385
354
+ C386
355
+ C387
356
+ C388
357
+ C389
358
+ C390
359
+ C391
360
+ C392
361
+ C393
362
+ C394
363
+ C395
364
+ C396
365
+ C397
366
+ C398
367
+ C399
368
+ C400
369
+ C401
370
+ C402
371
+ C403
372
+ C404
373
+ C406
374
+ C407
375
+ C408
376
+ C409
377
+ C410
378
+ C411
379
+ C412
380
+ C413
381
+ C414
382
+ C415
383
+ C416
384
+ C417
385
+ C418
386
+ C419
387
+ C420
388
+ C421
389
+ C422
390
+ C423
391
+ C424
392
+ C425
393
+ C426
394
+ C427
395
+ C428
396
+ C429
397
+ C430
398
+ C431
399
+ C432
400
+ C433
proteome_meta.py ADDED
@@ -0,0 +1,104 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import collections
2
+
3
+ # Global profiles
4
+ global_profiles_dict = {"Detectability": "detectability", "Promiscuity": "promiscuity"}
5
+
6
+ # Task suffixes
7
+ task_suf = collections.OrderedDict()
8
+
9
+ task_suf["_val_log2fc"] = "Log2FC"
10
+ task_suf["_gauss"] = "Gaussianized score"
11
+
12
+ task_suf["_bin_hit"] = "Protein hits (custom cutoffs)"
13
+ task_suf["_bin_50"] = "Top 50 proteins"
14
+ task_suf["_bin_100"] = "Top 100 proteins"
15
+ task_suf["_bin_250"] = "Top 250 proteins"
16
+ task_suf["_bin_500"] = "Top 500 proteins"
17
+ task_suf["_top_50"] = "Top 50 proteins"
18
+ task_suf["_top_100"] = "Top 100 proteins"
19
+ task_suf["_top_250"] = "Top 250 proteins"
20
+ task_suf["_top_500"] = "Top 500 proteins"
21
+ task_suf["_bottom_50"] = "Bottom 50 proteins"
22
+ task_suf["_bottom_100"] = "Bottom 100 proteins"
23
+ task_suf["_bottom_250"] = "Bottom 250 proteins"
24
+ task_suf["_bottom_500"] = "Bottom 500 proteins"
25
+
26
+ # Annotation type dict
27
+ annotation_type_dict = {
28
+ "Sequence": [
29
+ "InterPro Superfamily",
30
+ "InterPro Family",
31
+ "InterPro Domain",
32
+ "InterPro Active site",
33
+ "InterPro Binding site",
34
+ "InterPro Conserved site",
35
+ "InterPro Repeat",
36
+ "InterPro PTM"
37
+ ],
38
+ "Functions": [
39
+ "Panther Protein class",
40
+ "Protein Atlas Protein class",
41
+ "Protein Atlas Molecular function",
42
+ "GO Molecular function",
43
+ ],
44
+ "Processes and pathways": [
45
+ "Reactome Pathways",
46
+ "KEGG Pathways",
47
+ "WikiPathways",
48
+ "Protein Atlas Biological process",
49
+ "GO Biological process",
50
+ ],
51
+ "Localization": [
52
+ "Protein Atlas Subcellular",
53
+ "GO Cellular component",
54
+ "OpenCell Localization",
55
+ ],
56
+ "Drugs and Diseases": [
57
+ "Pharos IDG Category",
58
+ "Pharos Drug Target Ontology",
59
+ "Protein Atlas Disease involvement",
60
+ "Human Phenotype Ontology",
61
+ ],
62
+ }
63
+
64
+ # Annotation dict
65
+ annotation_dict = {
66
+ # Sequence
67
+ "InterPro Superfamily": "interpro_homologous_superfamily",
68
+ "InterPro Family": "interpro_family",
69
+ "InterPro Domain": "interpro_domain",
70
+ "InterPro Active site": "interpro_active_site",
71
+ "InterPro Binding site": "interpro_binding_site",
72
+ "InterPro Conserved site": "interpro_conserved_site",
73
+ "InterPro Repeat": "interpro_repeat",
74
+ "InterPro PTM": "interpro_ptm",
75
+ # Functions
76
+ "Panther Protein class": "panther_protein_class",
77
+ "Protein Atlas Protein class": "protein_atlas_protein_class",
78
+ "Protein Atlas Molecular function": "protein_atlas_molecular_function",
79
+ "GO Molecular function": "msigdb_gomf",
80
+ # Pathways
81
+ "Reactome Pathways": "msigdb_reactome",
82
+ "KEGG Pathways": "msigdb_kegg",
83
+ "WikiPathways": "msigdb_wp",
84
+ "Protein Atlas Biological process": "protein_atlas_biological_process",
85
+ "GO Biological process": "msigdb_gobp",
86
+ # Localization
87
+ "Protein Atlas Subcellular": "protein_atlas_subcellular_location_all", # _main?
88
+ "GO Cellular component": "msigdb_gocc",
89
+ "OpenCell Localization": "opencell_localization",
90
+ # Drugs and Diseases
91
+ "Pharos IDG Category": "pharos_protein_category",
92
+ "Pharos Drug Target Ontology": "pharos_dto",
93
+ "Protein Atlas Disease involvement": "protein_atlas_disease_involvement",
94
+ "Human Phenotype Ontology": "msigdb_hp",
95
+ }
96
+
97
+ # Universe dict
98
+ universe_dict = {
99
+ "Human Proteome": "human_proteome",
100
+ "HEK293T Core": "hek293t_core",
101
+ "Bind Degs Detected": "cemm_detected",
102
+ "Bind Degs Enriched": "cemm_enriched",
103
+ "Pulldown": "pulldown",
104
+ }