tanu/LSHTM_analysis
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

added untracked files in scripts and dynamut

Browse source
This commit is contained in:
Tanushree Tunstall 2022-01-04 12:26:54 +00:00
parent 00b84ccb1c
commit 3ab6a3dbc1
6 changed files with 1187 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

817
dynamut/katg_mcsm_formatted_snps_chain.csv Normal file
View file

@ -0,0 +1,817 @@
A G24V
A K27I
A K27E
A Y28L
A Y28H
A P29S
A V30A
A G32S
A G33S
A G34V
A G34A
A Q36P
A Q36H
A D37G
A P40T
A L43R
A L43P
A K46N
A V47I
A L48P
A L48R
A P52S
A D56H
A P57S
A A61S
A F62L
A D63G
A Y64C
A A65T
A A66T
A V68G
A I71F
A I71S
A V73A
A V73G
A A75P
A L76P
A T77R
A R78P
A R78G
A E81V
A E82D
A V83L
A V83G
A M84I
A M84T
A M84L
A T85A
A T85P
A T86P
A T86N
A S87L
A Q88P
A Q88E
A P89D
A W90R
A W90C
A W91G
A W91R
A W91L
A W91S
A P92T
A A93G
A A93D
A A93T
A D94N
A Y95F
A Y95S
A H97N
A H97P
A H97S
A Y98C
A Y98D
A Y98N
A G99R
A G99E
A P100T
A L101F
A L101M
A F102M
A F102S
A F102I
A I103N
A I103V
A I103T
A R104Q
A R104W
A M105I
A A106S
A A106V
A A106T
A A106R
A A106G
A A109T
A A109V
A A109S
A A109D
A A110V
A A110T
A G111D
A T112I
A Y113C
A I115V
A I115S
A I115T
A H116T
A H116E
A H116L
A H116G
A H116A
A H116Q
A H116F
A H116S
A H116P
A D117E
A G120S
A G121A
A G121S
A A122G
A A122D
A A122T
A A122V
A G123R
A G123E
A G124A
A G124Q
A G124D
A G124S
A G124H
A G124E
A G124R
A G124T
A G125D
A G125S
A M126Q
A M126I
A M126A
A M126L
A M126S
A Q127P
A R128Q
A R128L
A R128G
A R128W
A F129S
A A130E
A P131Q
A P131A
A P131L
A P131S
A L132R
A N133S
A N133D
A S134R
A W135S
A P136L
A N138S
A N138H
A N138D
A A139V
A A139P
A A139G
A S140N
A S140G
A S140I
A L141S
A L141F
A L141I
A L141V
A D142G
A D142N
A K143N
A K143E
A A144T
A A144V
A R145H
A R145C
A R145S
A R146L
A L148I
A W149R
A W149L
A W149G
A W149C
A V151L
A V151I
A K152E
A K152T
A K153Q
A Y155C
A Y155S
A Y155H
A G156D
A G156S
A K157N
A K157R
A K157Q
A K158S
A K158N
A L159I
A L159F
A L159P
A W161C
A W161R
A A162V
A A162E
A A162T
A D163N
A D163A
A L164R
A I165M
A I165L
A I165Y
A I165T
A V166I
A V166T
A F167S
A F167L
A F167C
A A168V
A A168T
A A168G
A G169S
A N170K
A C171V
A C171G
A A172T
A A172V
A L173R
A M176T
A M176I
A F178I
A F178S
A K179E
A T180M
A T180K
A G182R
A G182E
A F183L
A F183S
A G184D
A G184A
A G184C
A G186A
A G186S
A G186D
A R187P
A D189N
A D189G
A D189A
A D189Y
A W191R
A W191G
A E192A
A E192D
A D194N
A E195K
A V196G
A Y197D
A W204S
A L205R
A G206R
A E208K
A R209C
A S211N
A S211T
A K213E
A K213N
A R214L
A D215H
A D215E
A N218S
A P219L
A A222T
A Q224R
A M225V
A I228L
A N231K
A P232S
A P232R
A P232T
A P232A
A E233G
A E233Q
A G234R
A N236D
A G237A
A G237D
A P241H
A M242V
A M242T
A M242I
A A243T
A A244G
A V246R
A V246G
A I248T
A R249G
A R249C
A R249H
A T251K
A T251M
A F252L
A R253G
A R253W
A R254S
A R254C
A R254H
A R254L
A A256T
A A256V
A A256G
A M257I
A M257T
A M257V
A D259G
A D259E
A D259Y
A V260I
A V260E
A T262P
A A264V
A A264T
A V267A
A G268S
A G269S
A G269D
A T271P
A T271S
A T271I
A T271A
A F272L
A F272S
A F272V
A G273R
A G273C
A T275P
A T275A
A H276Q
A G277S
A G279D
A P280S
A P280Q
A A281V
A A281G
A A281T
A D282G
A G285C
A G285S
A G285V
A G285D
A G285A
A P286L
A P288H
A P288L
A E289A
A E289K
A A290V
A A290P
A A291D
A P292A
A Q295A
A Q295P
A Q295E
A M296V
A M296T
A G297V
A G297L
A L298S
A G299S
A G299C
A G299V
A G299A
A G299D
A W300S
A W300G
A W300R
A W300C
A S302R
A S302T
A G305C
A G305A
A T306A
A T306S
A G307R
A T308P
A T308S
A T308K
A T308A
A T308V
A T308I
A D311G
A A312P
A A312E
A A312V
A T314S
A T314N
A T314A
A S315T
A S315N
A S315I
A S315G
A S315R
A I317L
A I317V
A I317T
A E318K
A V320L
A V320A
A T322A
A T322M
A N323P
A N323S
A N323H
A T324N
A T324P
A T324S
A T324L
A P325S
A P325T
A T326P
A T326M
A K327T
A W328L
A W328S
A W328R
A W328C
A D329A
A D329E
A D329H
A S331T
A S331I
A S331R
A L333F
A L333C
A E334K
A I335V
A I335T
A I335N
A L336M
A Y337C
A Y337H
A Y337F
A Y337S
A G338S
A Y339N
A Y339C
A Y339S
A E340D
A E342G
A T344L
A T344K
A T344S
A T344M
A A348V
A A348G
A G349D
A Q352Y
A Y353H
A Y353F
A T354I
A D357H
A I364N
A D366N
A P367L
A F368L
A S374A
A S374P
A L378P
A L378M
A A379V
A A379T
A T380S
A T380P
A T380I
A T380A
A T380N
A D381A
A L382I
A L382R
A S383W
A S383A
A L384R
A R385P
A V386M
A V386E
A D387N
A Y390C
A R392W
A T394P
A T394M
A T394A
A R395C
A L398R
A E399D
A E399K
A H400Y
A H400P
A E402A
A E402K
A L404W
A D406A
A D406E
A D406G
A E407A
A E407K
A F408Y
A F408S
A F408L
A F408V
A A411D
A Y413C
A Y413F
A Y413H
A Y413S
A K414R
A I416M
A I416T
A I416L
A I416V
A D419H
A D419G
A D419Y
A D419V
A P422H
A P422L
A V423I
A A424V
A A424G
A R425K
A L427P
A L427R
A L427F
A L430A
A P432L
A P432T
A K433T
A Q434P
A L437R
A W438G
A Q439K
A Q439H
A Q439R
A Q439T
A D440G
A P441L
A V442L
A V442A
A V445I
A S446N
A D448A
A D448E
A V450I
A V450A
A G451D
A E452Q
A I455L
A L458H
A K459T
A S460N
A Q461P
A Q461R
A Q461E
A I462S
A R463L
A R463W
A S465P
A T468P
A V469L
A V469I
A Q471R
A V473L
A V473F
A S474Q
A T475I
A T475A
A A476E
A A476V
A A478R
A A479P
A A479G
A A479V
A A479Q
A A480Q
A A480S
A S481A
A S481L
A S482T
A F483L
A R484H
A R484G
A K488E
A R489C
A G490D
A G490C
A G490S
A G491S
A A492V
A A492D
A N493K
A G494S
A G494A
A G495S
A G495A
A G495C
A R496L
A R496C
A R498S
A P501S
A V503A
A V503S
A W505L
A V507I
A N508D
A D509E
A D509N
A P510A
A D511N
A D513N
A L514P
A L514V
A R515H
A K516R
A R519H
A T520A
A L521P
A E522K
A E523D
A Q525P
A Q525A
A Q525K
A Q525S
A E526D
A S527L
A N529T
A A532P
A A532V
A P533L
A G534A
A G534R
A K537E
A V538A
A F540S
A A541T
A D542E
A L546F
A C549S
A A550D
A A551S
A A555P
A A556S
A K557N
A G560R
A G560A
A G560S
A H561R
A N562H
A V565G
A P566L
A F567S
A F567L
A F567V
A T568P
A P569L
A G570F
A R571L
A A574V
A T579A
A T579S
A S583P
A F584V
A V586M
A L587R
A L587P
A E588G
A A591T
A G593C
A F594I
A F594L
A N596S
A Y597H
A Y597S
A Y597D
A L598F
A L598R
A G599R
A K600Q
A N602D
A P603L
A P605S
A A606P
A A606T
A E607D
A Y608D
A M609T
A L611R
A D612G
A A614T
A A614G
A A614E
A L616S
A T618M
A S620T
A A621T
A A621D
A M624V
A M624K
A M624I
A T625A
A T625K
A L627P
A V628I
A G629D
A G629C
A G630R
A G630V
A V633A
A V633I
A L634I
A A636T
A N637D
A N637H
A N637K
A Y638C
A Y638H
A G644D
A G644S
A G644V
A E648D
A A649T
A A649G
A S650F
A S650P
A E651D
A L653Q
A T654S
A N655D
A F657S
A F657L
A N660D
A L661M
A L662V
A D663G
A D663Y
A I666V
A T667P
A T667I
A W668C
A W668L
A A673V
A D675Y
A D675G
A D675H
A T677P
A Y678C
A Q679E
A Q679Y
A G680D
A K681Q
A K681T
A S684R
A K686E
A W689G
A W689R
A T690I
A T690P
A G691D
A S692R
A R693C
A R693H
A D695A
A L696Q
A L696P
A V697A
A F698V
A G699E
A G699V
A S700P
A S700F
A E703Q
A L704W
A L704S
A R705L
A R705G
A R705W
A L707R
A L707F
A E709A
A E709G
A V710I
A V710A
A Y711D
A A713S
A D714E
A D714N
A D714G
A P718S
A F720S
A D723N
A D723A
A A726T
A A727S
A A727T
A W728R
A D729N
A D729V
A D729G
A D729T
A V731M
A V731A
A N733S
A L734R
A D735A
A R736K
A R736S
A V739M
A R740S
1 A G24V
2 A K27I
3 A K27E
4 A Y28L
5 A Y28H
6 A P29S
7 A V30A
8 A G32S
9 A G33S
10 A G34V
11 A G34A
12 A Q36P
13 A Q36H
14 A D37G
15 A P40T
16 A L43R
17 A L43P
18 A K46N
19 A V47I
20 A L48P
21 A L48R
22 A P52S
23 A D56H
24 A P57S
25 A A61S
26 A F62L
27 A D63G
28 A Y64C
29 A A65T
30 A A66T
31 A V68G
32 A I71F
33 A I71S
34 A V73A
35 A V73G
36 A A75P
37 A L76P
38 A T77R
39 A R78P
40 A R78G
41 A E81V
42 A E82D
43 A V83L
44 A V83G
45 A M84I
46 A M84T
47 A M84L
48 A T85A
49 A T85P
50 A T86P
51 A T86N
52 A S87L
53 A Q88P
54 A Q88E
55 A P89D
56 A W90R
57 A W90C
58 A W91G
59 A W91R
60 A W91L
61 A W91S
62 A P92T
63 A A93G
64 A A93D
65 A A93T
66 A D94N
67 A Y95F
68 A Y95S
69 A H97N
70 A H97P
71 A H97S
72 A Y98C
73 A Y98D
74 A Y98N
75 A G99R
76 A G99E
77 A P100T
78 A L101F
79 A L101M
80 A F102M
81 A F102S
82 A F102I
83 A I103N
84 A I103V
85 A I103T
86 A R104Q
87 A R104W
88 A M105I
89 A A106S
90 A A106V
91 A A106T
92 A A106R
93 A A106G
94 A A109T
95 A A109V
96 A A109S
97 A A109D
98 A A110V
99 A A110T
100 A G111D
101 A T112I
102 A Y113C
103 A I115V
104 A I115S
105 A I115T
106 A H116T
107 A H116E
108 A H116L
109 A H116G
110 A H116A
111 A H116Q
112 A H116F
113 A H116S
114 A H116P
115 A D117E
116 A G120S
117 A G121A
118 A G121S
119 A A122G
120 A A122D
121 A A122T
122 A A122V
123 A G123R
124 A G123E
125 A G124A
126 A G124Q
127 A G124D
128 A G124S
129 A G124H
130 A G124E
131 A G124R
132 A G124T
133 A G125D
134 A G125S
135 A M126Q
136 A M126I
137 A M126A
138 A M126L
139 A M126S
140 A Q127P
141 A R128Q
142 A R128L
143 A R128G
144 A R128W
145 A F129S
146 A A130E
147 A P131Q
148 A P131A
149 A P131L
150 A P131S
151 A L132R
152 A N133S
153 A N133D
154 A S134R
155 A W135S
156 A P136L
157 A N138S
158 A N138H
159 A N138D
160 A A139V
161 A A139P
162 A A139G
163 A S140N
164 A S140G
165 A S140I
166 A L141S
167 A L141F
168 A L141I
169 A L141V
170 A D142G
171 A D142N
172 A K143N
173 A K143E
174 A A144T
175 A A144V
176 A R145H
177 A R145C
178 A R145S
179 A R146L
180 A L148I
181 A W149R
182 A W149L
183 A W149G
184 A W149C
185 A V151L
186 A V151I
187 A K152E
188 A K152T
189 A K153Q
190 A Y155C
191 A Y155S
192 A Y155H
193 A G156D
194 A G156S
195 A K157N
196 A K157R
197 A K157Q
198 A K158S
199 A K158N
200 A L159I
201 A L159F
202 A L159P
203 A W161C
204 A W161R
205 A A162V
206 A A162E
207 A A162T
208 A D163N
209 A D163A
210 A L164R
211 A I165M
212 A I165L
213 A I165Y
214 A I165T
215 A V166I
216 A V166T
217 A F167S
218 A F167L
219 A F167C
220 A A168V
221 A A168T
222 A A168G
223 A G169S
224 A N170K
225 A C171V
226 A C171G
227 A A172T
228 A A172V
229 A L173R
230 A M176T
231 A M176I
232 A F178I
233 A F178S
234 A K179E
235 A T180M
236 A T180K
237 A G182R
238 A G182E
239 A F183L
240 A F183S
241 A G184D
242 A G184A
243 A G184C
244 A G186A
245 A G186S
246 A G186D
247 A R187P
248 A D189N
249 A D189G
250 A D189A
251 A D189Y
252 A W191R
253 A W191G
254 A E192A
255 A E192D
256 A D194N
257 A E195K
258 A V196G
259 A Y197D
260 A W204S
261 A L205R
262 A G206R
263 A E208K
264 A R209C
265 A S211N
266 A S211T
267 A K213E
268 A K213N
269 A R214L
270 A D215H
271 A D215E
272 A N218S
273 A P219L
274 A A222T
275 A Q224R
276 A M225V
277 A I228L
278 A N231K
279 A P232S
280 A P232R
281 A P232T
282 A P232A
283 A E233G
284 A E233Q
285 A G234R
286 A N236D
287 A G237A
288 A G237D
289 A P241H
290 A M242V
291 A M242T
292 A M242I
293 A A243T
294 A A244G
295 A V246R
296 A V246G
297 A I248T
298 A R249G
299 A R249C
300 A R249H
301 A T251K
302 A T251M
303 A F252L
304 A R253G
305 A R253W
306 A R254S
307 A R254C
308 A R254H
309 A R254L
310 A A256T
311 A A256V
312 A A256G
313 A M257I
314 A M257T
315 A M257V
316 A D259G
317 A D259E
318 A D259Y
319 A V260I
320 A V260E
321 A T262P
322 A A264V
323 A A264T
324 A V267A
325 A G268S
326 A G269S
327 A G269D
328 A T271P
329 A T271S
330 A T271I
331 A T271A
332 A F272L
333 A F272S
334 A F272V
335 A G273R
336 A G273C
337 A T275P
338 A T275A
339 A H276Q
340 A G277S
341 A G279D
342 A P280S
343 A P280Q
344 A A281V
345 A A281G
346 A A281T
347 A D282G
348 A G285C
349 A G285S
350 A G285V
351 A G285D
352 A G285A
353 A P286L
354 A P288H
355 A P288L
356 A E289A
357 A E289K
358 A A290V
359 A A290P
360 A A291D
361 A P292A
362 A Q295A
363 A Q295P
364 A Q295E
365 A M296V
366 A M296T
367 A G297V
368 A G297L
369 A L298S
370 A G299S
371 A G299C
372 A G299V
373 A G299A
374 A G299D
375 A W300S
376 A W300G
377 A W300R
378 A W300C
379 A S302R
380 A S302T
381 A G305C
382 A G305A
383 A T306A
384 A T306S
385 A G307R
386 A T308P
387 A T308S
388 A T308K
389 A T308A
390 A T308V
391 A T308I
392 A D311G
393 A A312P
394 A A312E
395 A A312V
396 A T314S
397 A T314N
398 A T314A
399 A S315T
400 A S315N
401 A S315I
402 A S315G
403 A S315R
404 A I317L
405 A I317V
406 A I317T
407 A E318K
408 A V320L
409 A V320A
410 A T322A
411 A T322M
412 A N323P
413 A N323S
414 A N323H
415 A T324N
416 A T324P
417 A T324S
418 A T324L
419 A P325S
420 A P325T
421 A T326P
422 A T326M
423 A K327T
424 A W328L
425 A W328S
426 A W328R
427 A W328C
428 A D329A
429 A D329E
430 A D329H
431 A S331T
432 A S331I
433 A S331R
434 A L333F
435 A L333C
436 A E334K
437 A I335V
438 A I335T
439 A I335N
440 A L336M
441 A Y337C
442 A Y337H
443 A Y337F
444 A Y337S
445 A G338S
446 A Y339N
447 A Y339C
448 A Y339S
449 A E340D
450 A E342G
451 A T344L
452 A T344K
453 A T344S
454 A T344M
455 A A348V
456 A A348G
457 A G349D
458 A Q352Y
459 A Y353H
460 A Y353F
461 A T354I
462 A D357H
463 A I364N
464 A D366N
465 A P367L
466 A F368L
467 A S374A
468 A S374P
469 A L378P
470 A L378M
471 A A379V
472 A A379T
473 A T380S
474 A T380P
475 A T380I
476 A T380A
477 A T380N
478 A D381A
479 A L382I
480 A L382R
481 A S383W
482 A S383A
483 A L384R
484 A R385P
485 A V386M
486 A V386E
487 A D387N
488 A Y390C
489 A R392W
490 A T394P
491 A T394M
492 A T394A
493 A R395C
494 A L398R
495 A E399D
496 A E399K
497 A H400Y
498 A H400P
499 A E402A
500 A E402K
501 A L404W
502 A D406A
503 A D406E
504 A D406G
505 A E407A
506 A E407K
507 A F408Y
508 A F408S
509 A F408L
510 A F408V
511 A A411D
512 A Y413C
513 A Y413F
514 A Y413H
515 A Y413S
516 A K414R
517 A I416M
518 A I416T
519 A I416L
520 A I416V
521 A D419H
522 A D419G
523 A D419Y
524 A D419V
525 A P422H
526 A P422L
527 A V423I
528 A A424V
529 A A424G
530 A R425K
531 A L427P
532 A L427R
533 A L427F
534 A L430A
535 A P432L
536 A P432T
537 A K433T
538 A Q434P
539 A L437R
540 A W438G
541 A Q439K
542 A Q439H
543 A Q439R
544 A Q439T
545 A D440G
546 A P441L
547 A V442L
548 A V442A
549 A V445I
550 A S446N
551 A D448A
552 A D448E
553 A V450I
554 A V450A
555 A G451D
556 A E452Q
557 A I455L
558 A L458H
559 A K459T
560 A S460N
561 A Q461P
562 A Q461R
563 A Q461E
564 A I462S
565 A R463L
566 A R463W
567 A S465P
568 A T468P
569 A V469L
570 A V469I
571 A Q471R
572 A V473L
573 A V473F
574 A S474Q
575 A T475I
576 A T475A
577 A A476E
578 A A476V
579 A A478R
580 A A479P
581 A A479G
582 A A479V
583 A A479Q
584 A A480Q
585 A A480S
586 A S481A
587 A S481L
588 A S482T
589 A F483L
590 A R484H
591 A R484G
592 A K488E
593 A R489C
594 A G490D
595 A G490C
596 A G490S
597 A G491S
598 A A492V
599 A A492D
600 A N493K
601 A G494S
602 A G494A
603 A G495S
604 A G495A
605 A G495C
606 A R496L
607 A R496C
608 A R498S
609 A P501S
610 A V503A
611 A V503S
612 A W505L
613 A V507I
614 A N508D
615 A D509E
616 A D509N
617 A P510A
618 A D511N
619 A D513N
620 A L514P
621 A L514V
622 A R515H
623 A K516R
624 A R519H
625 A T520A
626 A L521P
627 A E522K
628 A E523D
629 A Q525P
630 A Q525A
631 A Q525K
632 A Q525S
633 A E526D
634 A S527L
635 A N529T
636 A A532P
637 A A532V
638 A P533L
639 A G534A
640 A G534R
641 A K537E
642 A V538A
643 A F540S
644 A A541T
645 A D542E
646 A L546F
647 A C549S
648 A A550D
649 A A551S
650 A A555P
651 A A556S
652 A K557N
653 A G560R
654 A G560A
655 A G560S
656 A H561R
657 A N562H
658 A V565G
659 A P566L
660 A F567S
661 A F567L
662 A F567V
663 A T568P
664 A P569L
665 A G570F
666 A R571L
667 A A574V
668 A T579A
669 A T579S
670 A S583P
671 A F584V
672 A V586M
673 A L587R
674 A L587P
675 A E588G
676 A A591T
677 A G593C
678 A F594I
679 A F594L
680 A N596S
681 A Y597H
682 A Y597S
683 A Y597D
684 A L598F
685 A L598R
686 A G599R
687 A K600Q
688 A N602D
689 A P603L
690 A P605S
691 A A606P
692 A A606T
693 A E607D
694 A Y608D
695 A M609T
696 A L611R
697 A D612G
698 A A614T
699 A A614G
700 A A614E
701 A L616S
702 A T618M
703 A S620T
704 A A621T
705 A A621D
706 A M624V
707 A M624K
708 A M624I
709 A T625A
710 A T625K
711 A L627P
712 A V628I
713 A G629D
714 A G629C
715 A G630R
716 A G630V
717 A V633A
718 A V633I
719 A L634I
720 A A636T
721 A N637D
722 A N637H
723 A N637K
724 A Y638C
725 A Y638H
726 A G644D
727 A G644S
728 A G644V
729 A E648D
730 A A649T
731 A A649G
732 A S650F
733 A S650P
734 A E651D
735 A L653Q
736 A T654S
737 A N655D
738 A F657S
739 A F657L
740 A N660D
741 A L661M
742 A L662V
743 A D663G
744 A D663Y
745 A I666V
746 A T667P
747 A T667I
748 A W668C
749 A W668L
750 A A673V
751 A D675Y
752 A D675G
753 A D675H
754 A T677P
755 A Y678C
756 A Q679E
757 A Q679Y
758 A G680D
759 A K681Q
760 A K681T
761 A S684R
762 A K686E
763 A W689G
764 A W689R
765 A T690I
766 A T690P
767 A G691D
768 A S692R
769 A R693C
770 A R693H
771 A D695A
772 A L696Q
773 A L696P
774 A V697A
775 A F698V
776 A G699E
777 A G699V
778 A S700P
779 A S700F
780 A E703Q
781 A L704W
782 A L704S
783 A R705L
784 A R705G
785 A R705W
786 A L707R
787 A L707F
788 A E709A
789 A E709G
790 A V710I
791 A V710A
792 A Y711D
793 A A713S
794 A D714E
795 A D714N
796 A D714G
797 A P718S
798 A F720S
799 A D723N
800 A D723A
801 A A726T
802 A A727S
803 A A727T
804 A W728R
805 A D729N
806 A D729V
807 A D729G
808 A D729T
809 A V731M
810 A V731A
811 A N733S
812 A L734R
813 A D735A
814 A R736K
815 A R736S
816 A V739M
817 A R740S

63
scripts/functions/test_aa_prop_bp.R Normal file
View file

@ -0,0 +1,63 @@
#!/usr/bin/env Rscript
library(ggplot2)
library(tidyverse)
library(data.table)
setwd("~/git/LSHTM_analysis/scripts/functions/")
getwd()
#############################################################
#===========================================
# load functions, data, dirs, hardocded vars
# that will be used in testing the functions
#===========================================
source("plotting_data.R")
infile = "/home/tanu/git/Data/streptomycin/output/"
pd_df = plotting_data(infile)
my_df = pd_df[[1]]
my_df_u = pd_df[[2]]
my_df_u_lig = pd_df[[3]]
dup_muts = pd_df[[4]]
source("../plotting_globals.R")
drug = "streptomycin"
gene = "gid"
import_dirs(drug, gene)
#=====================
# functions to test
#=====================
source("stability_count_bp.R")
source("position_count_bp.R")
#################################################################
##############################################
# read a sample file containing muts and prop
###############################################
df<- read.csv(file.choose())
setDT(df)[, pos_count := .N, by = .(position)]
foo = data.frame(df$position, df$pos_count)
#snpsBYpos_df <- df %>%
# group_by(position) %>%
# summarize(snpsBYpos = mean(pos_count))
# subset df without duplicates for position
df2 = df[!duplicated(df$position)]
##################################################################
# ---------------------------------------
# barplot for nssnps, coloured by aa prop
# ---------------------------------------
pos_colname = "position"
aa_prop_colname = "mut_prop_water"
aa_prop_colours = c("black", "blue")
my_legname = "aa_prop: water"
# call function
aa_prop_bp(plotdf = df
, position_colname = pos_colname
, fill_colname = aa_prop_colname
, fill_colours = aa_prop_cols
, leg_name = my_legname)
#===============================================================

59
scripts/functions/test_af_or_calcs.R Normal file
View file

@ -0,0 +1,59 @@
#!/usr/bin/env Rscript
#########################################################
# TASK: To calculate Allele Frequency and
# Odds Ratio from master data
#########################################################
# load libraries
#source("Header_TT.R")
require("getopt", quietly = TRUE) # cmd parse arguments
# working dir and loading libraries
getwd()
setwd("~/git/LSHTM_analysis/scripts/functions/")
getwd()
# load functions
source("plotting_globals.R")
source("mychisq_or.R")
source("myaf_or_calcs.R")
# cmd options + sensible defaults
drug = "streptomycin"
gene = "gid"
# call function
import_dirs(drug, gene)
# input file 1: master data
#in_filename_master = 'original_tanushree_data_v2.csv' #19K
in_filename_master = 'mtb_gwas_meta_v6.csv' #35k
infile_master = paste0(datadir, in_filename_master)
cat(paste0('Reading infile1: raw data', ' ', infile_master) )
# input file 2: gene associated meta data file to extract valid snps and add calcs to.
# This is outfile_metadata from data_extraction.py
in_filename_metadata = paste0(tolower(gene), '_metadata.csv')
infile_metadata = paste0(outdir, '/', in_filename_metadata)
cat(paste0('Reading input file 2 i.e gene associated metadata:', infile_metadata))
# out_filename_af_or = paste0(tolower(gene), '_meta_data_with_AF_OR.csv')
out_filename_af_or = paste0(tolower(gene), '_af_or.csv')
outfile_af_or = paste0(outdir, '/', out_filename_af_or)
cat(paste0('Output file with full path:', outfile_af_or))
cat("master data:", infile_master)
cat("gene data:", infile_metadata)
dr_muts_col # comes from global (dr_mutations_<drug>)
other_muts_col # comes from global (other_mutations_<drug>)
#################################################
my_afor ( infile_master
, infile_metadata
, outfile = outfile_af_or
#, outfile = "FOO_TEST.csv"
, drug
, gene
, idcol = "id"
, dr_muts_col
, other_muts_col
)

113
scripts/functions/test_bp.R Normal file
View file

@ -0,0 +1,113 @@
#!/usr/bin/env Rscript
setwd("~/git/LSHTM_analysis/scripts/functions/")
getwd()
#############################################################
#===========================================
# load functions, data, dirs, hardocded vars
# that will be used in testing the functions
#===========================================
drug = "streptomycin"
gene = "gid"
source("plotting_data.R")
infile = paste0("~/git/Data/", drug, "/output/", gene, "_comb_stab_struc_params.csv")
infile_df = read.csv(infile)
lig_dist = 5
pd_df = plotting_data(infile_df
, lig_dist_colname = 'ligand_distance'
, lig_dist_cutoff = lig_dist)
my_df = pd_df[[1]]
my_df_u = pd_df[[2]]
my_df_u_lig = pd_df[[3]]
dup_muts = pd_df[[4]]
#=====================
# functions to test
#=====================
source("stability_count_bp.R")
source("position_count_bp.R")
##################################################################
# ------------------------------
# barplot for mscm stability
# ------------------------------
basic_bp_duet = paste0(tolower(gene), "_basic_barplot_PS.svg")
plot_basic_bp_duet = paste0(plotdir,"/", basic_bp_duet)
svg(plot_basic_bp_duet)
print(paste0("plot filename:", basic_bp_duet))
# function only
stability_count_bp(plotdf = my_df_u
, df_colname = "duet_outcome"
, leg_title = "DUET outcome"
, label_categories = c("Destabilising", "Stabilising")
, leg_position = "top")
dev.off()
# ------------------------------
# barplot for ligand affinity
# ------------------------------
basic_bp_ligand = paste0(tolower(gene), "_basic_barplot_LIG.svg")
plot_basic_bp_ligand = paste0(plotdir, "/", basic_bp_ligand)
svg(plot_basic_bp_ligand)
print(paste0("plot filename:", basic_bp_ligand))
# function only
lig_dist = 10
stability_count_bp(plotdf = my_df_u_lig
, df_colname = "ligand_outcome"
, leg_title = "Ligand outcome"
, yaxis_title = paste0("Number of nsSNPs\nLigand dist: <", lig_dist, "\u212b")
#, bp_plot_title = "Sites < 10 Ang of ligand"
)
dev.off()
# ------------------------------
# barplot for foldX
# ------------------------------
basic_bp_foldx = paste0(tolower(gene), "_basic_barplot_foldx.svg")
plot_basic_bp_foldx = paste0(plotdir,"/", basic_bp_foldx)
svg(plot_basic_bp_foldx)
print(paste0("plot filename:", plot_basic_bp_foldx))
stability_count_bp(plotdf = my_df_u
, df_colname = "foldx_outcome"
, leg_title = "FoldX outcome")
dev.off()
#===============================================================
# ------------------------------
# barplot for nssnp site count: all
# ------------------------------
pos_count_duet = paste0(tolower(gene), "_position_count_PS.svg")
plot_pos_count_duet = paste0(plotdir, "/", pos_count_duet)
svg(plot_pos_count_duet)
print(paste0("plot filename:", plot_pos_count_duet))
# function only
site_snp_count_bp(plotdf = my_df_u
, df_colname = "position")
dev.off()
# ------------------------------
# barplot for nssnp site count: within 10 Ang
# ------------------------------
pos_count_ligand = paste0(tolower(gene), "_position_count_LIG.svg")
plot_pos_count_ligand = paste0(plotdir, "/", pos_count_ligand)
svg(plot_pos_count_ligand)
print(paste0("plot filename:", plot_pos_count_ligand))
# function only
site_snp_count_bp(plotdf = my_df_u_lig
, df_colname = "position")
dev.off()
#===============================================================

100
scripts/functions/test_combining_dfs_plotting.R Normal file
View file

@ -0,0 +1,100 @@
#!/usr/bin/env Rscript
# working dir and loading libraries
getwd()
setwd("~/git/LSHTM_analysis/scripts/functions/")
getwd()
# infile_params = paste0(outdir, "/" , tolower(gene), "_comb_afor.csv")
# infile_metadata = paste0(outdir, "/", tolower(gene), "_metadata")
#
#
# source("combining_dfs_plotting_func.R")
#
####################################################################
# in_file_params = "~/git/Data/streptomycin/output/gid_comb_afor.csv"
# in_file_metadata = "~/git/Data/streptomycin/output/gid_metadata.csv"
#
# all_plot_dfs = combining_dfs_plotting(df1_mcsm_comb = infile_params
# , df2_gene_metadata = infile_metadata
# , lig_dist_colname = 'ligand_distance'
# , lig_dist_cutoff = 10)
#
# merged_df2 = all_plot_dfs[[1]]
# merged_df3 = all_plot_dfs[[2]]
# merged_df2_comp = all_plot_dfs[[3]]
# merged_df3_comp = all_plot_dfs[[4]]
# merged_df2_lig = all_plot_dfs[[5]]
# merged_df3_lig = all_plot_dfs[[6]]
#
# bar_colnames = data.frame(colnames(merged_df2))
###########################################################
source("plotting_globals.R")
source("plotting_data.R")
source("combining_dfs_plotting.R")
#---------------------
# call: import_dirs()
#---------------------
gene = 'gid'
drug = 'streptomycin'
import_dirs(drug_name = drug, gene_name = gene)
#============================
# Input 1: plotting_data()
#============================
if (!exists("infile_params") && exists("gene")){
#if (!is.character(infile_params) && exists("gene")){
#in_filename_params = paste0(tolower(gene), "_all_params.csv")
in_filename_params = paste0(tolower(gene), "_comb_afor.csv") # part combined for gid
infile_params = paste0(outdir, "/", in_filename_params)
cat("\nInput file for mcsm comb data not specified, assuming filename: ", infile_params, "\n")
}
mcsm_comb_data = read.csv(infile_params, header = T)
#-------------------------------
# call function: plotting_data()
#-------------------------------
pd_df = plotting_data(df = mcsm_comb_data
, ligand_dist_colname = 'ligand_distance'
, lig_dist_cutoff = 10
my_df_u = pd_df[[2]]
#======================================
# Input 2: read <gene>_meta data.csv
#======================================
if (!exists("infile_metadata") && exists("gene")){
#if (!is.character(infile_params) && exists("gene")){{
in_filename_metadata = paste0(tolower(gene), "_metadata.csv") # part combined for gid
infile_metadata = paste0(outdir, "/", in_filename_metadata)
cat("\nInput file for gene metadata not specified, assuming filename: ", infile_metadata, "\n")
}
cat("\nReading meta data file:", infile_metadata)
gene_metadata <- read.csv(infile_metadata
, stringsAsFactors = F
, header = T)
#-----------------------------------------
# test function: combining_dfs_plotting()
#-----------------------------------------
all_plot_dfs = combining_dfs_plotting(my_df_u
, gene_metadata
, lig_dist_colname = 'ligand_distance'
, lig_dist_cutoff = 10)
merged_df2 = all_plot_dfs[[1]]
merged_df3 = all_plot_dfs[[2]]
merged_df2_comp = all_plot_dfs[[3]]
merged_df3_comp = all_plot_dfs[[4]]
merged_df2_lig = all_plot_dfs[[5]]
merged_df3_lig = all_plot_dfs[[6]]
merged_df2_comp_lig = all_plot_dfs[[7]]
merged_df3_comp_lig = all_plot_dfs[[8]]
########################################################################
# End of script
########################################################################

35
scripts/functions/test_plotting_data.R Normal file
View file

@ -0,0 +1,35 @@
#!/usr/bin/env Rscript
getwd()
setwd("~/git/LSHTM_analysis/scripts/functions/")
getwd()
#############################################################
#===========================================
# load functions, data, dirs, hardocded vars
# that will be used in testing the functions
#===========================================
source("plotting_globals.R")
drug = "streptomycin"
gene = "gid"
import_dirs(drug_name = drug, gene_name = gene)
#-------------------------------
# test function: plotting_data()
#-------------------------------
source("plotting_data.R")
infile_params = "/home/tanu/git/Data/streptomycin/output/gid_comb_stab_struc_params.csv"
mcsm_comb_data = read.csv(infile_params, header = T)
pd_df = plotting_data(df = mcsm_comb_data
, ligand_dist_colname = 'ligand_distance'
, lig_dist_cutoff = 10)
my_df = pd_df[[1]]
my_df_u = pd_df[[2]]
my_df_u_lig = pd_df[[3]]
dup_muts = pd_df[[4]]
########################################################################
# End of script
########################################################################