-
Notifications
You must be signed in to change notification settings - Fork 14
/
01R-Merge-WGS-RNA-data-decontaminate-batch-correct.R
executable file
·460 lines (395 loc) · 26.7 KB
/
01R-Merge-WGS-RNA-data-decontaminate-batch-correct.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
#-----------------------------------------------------------------------------
# 01R-Merge-WGS-RNA-data-decontaminate-batch-correct.R
# Copyright (c) 2021--, Greg Poore
# Purposes:
# - Merge WGS and RNA-Seq datasets
# - Decontaminate using the plate-center method
# - Measure batch effects using PVCA
# - Batch correct using Voom-SNM
#-----------------------------------------------------------------------------
#----------------------------------------------------------#
# Load environments
#----------------------------------------------------------#
require(doMC)
require(plyr)
require(dplyr)
require(tibble)
require(biomformat)
require(rhdf5)
require(ggpubr)
require(ggsci)
numCores <- detectCores()
registerDoMC(cores=numCores)
#----------------------------------------------------------#
# Rep200 fungal species identification
#----------------------------------------------------------#
rep200TaxSplit <- read.csv("Supporting_data/rep200_lineage_map_split.csv", stringsAsFactors = FALSE, row.names = 1)
rep200Kingdoms <- read.csv("Supporting_data/rep200_gOTU_kingdom_mapping.csv", stringsAsFactors = FALSE)
rep200Kingdoms_Fungi <- rep200Kingdoms[which(rep200Kingdoms$kingdom == "fungi"),]
rep200TaxSplit_Fungi <- rep200TaxSplit[rownames(rep200TaxSplit) %in% rep200Kingdoms_Fungi$genomeID,,drop=FALSE]
dim(rep200TaxSplit_Fungi) # 320 7
#----------------------------------------------------------#
# Microbial data import - WGS
#----------------------------------------------------------#
## Import metadata and read count data
metaQiita <- read.csv("Input_data/tcga_wgs_reprocess_qiita_metadata_2Apr21.csv", stringsAsFactors = FALSE, row.names = 1)
rep200Data_WGS_BIOM <- read_biom(biom_file = "Input_data/Qiita_results/rep200_TCGA_WGS_OGU_25Mar22.biom")
rep200Data_WGS <- t(as(biom_data(rep200Data_WGS_BIOM), "matrix"))
rownames(rep200Data_WGS) <- gsub("^11[0-9]+\\.","",rownames(rep200Data_WGS))
dim(rep200Data_WGS) # 4736 11585
# Check rowname overlap and subset metadata
sum(rownames(rep200Data_WGS) %in% rownames(metaQiita)) # 4736
metaQiitaWGS <- droplevels(metaQiita[rownames(rep200Data_WGS),])
## Extract only fungal features
rep200Data_WGS_Fungi <- rep200Data_WGS[,colnames(rep200Data_WGS) %in% rownames(rep200TaxSplit_Fungi)]
dim(rep200Data_WGS_Fungi) # 4736 318
# Extract HiSeq samples
metaQiitaWGS_HiSeq <- metaQiitaWGS %>% filter(platform_tcga == "Illumina HiSeq") %>% droplevels()
dim(metaQiitaWGS_HiSeq) # 4387 39
wgsSampleIDintersectHiSeq <- intersect(rownames(metaQiitaWGS_HiSeq), rownames(rep200Data_WGS))
metaQiitaWGS_HiSeq_Filt <- droplevels(metaQiitaWGS_HiSeq[wgsSampleIDintersectHiSeq,])
rep200Data_WGS_HiSeq <- rep200Data_WGS[wgsSampleIDintersectHiSeq,]
rep200Data_WGS_HiSeq_Fungi <- rep200Data_WGS_HiSeq[,colnames(rep200Data_WGS_HiSeq) %in% rownames(rep200TaxSplit_Fungi)]
dim(rep200Data_WGS_HiSeq_Fungi) # 4387 318
#----------------------------------------------------------#
# Microbial data import - RNA
#----------------------------------------------------------#
## Import metadata and read count data
metaQiitaRNA <- read.csv("Input_data/qiita_metadata_tcga_rna_reprocess_28Apr21.csv", stringsAsFactors = FALSE)
rownames(metaQiitaRNA) <- paste0("13767.",metaQiitaRNA$sample_name)
rep200Data_RNA_BIOM <- read_biom(biom_file = "Input_data/Qiita_results/rep200_TCGA_RNA_OGU_25Mar22.biom")
rep200Data_RNA <- t(as(biom_data(rep200Data_RNA_BIOM), "matrix"))
rownames(rep200Data_RNA) <- gsub("^11[0-9]+\\.","",rownames(rep200Data_RNA)) # Qiita IDs get appended to name; this removes them
dim(rep200Data_RNA) # 10776 11735
# Extract HiSeq samples
metaQiitaRNA_HiSeq <- metaQiitaRNA %>% filter(cgc_platform == "Illumina HiSeq") %>% droplevels()
dim(metaQiitaRNA_HiSeq) # 10701 41
rnaSampleIDintersectHiSeq <- intersect(rownames(metaQiitaRNA_HiSeq),
rownames(rep200Data_RNA))
metaQiitaRNA_HiSeq_Filt <- droplevels(metaQiitaRNA_HiSeq[rnaSampleIDintersectHiSeq,])
rep200Data_RNA_HiSeq <- rep200Data_RNA[rnaSampleIDintersectHiSeq,]
rep200Data_RNA_HiSeq_Fungi <- rep200Data_RNA_HiSeq[,colnames(rep200Data_RNA_HiSeq) %in% rownames(rep200TaxSplit_Fungi)]
dim(rep200Data_RNA_HiSeq_Fungi) # 10701 319
#-----------------------------------------------#
# Combine metadata data - all sequencing platforms
# To be used in "02-Calculate-fungi-vs-bacteria-read-distributions.R"
#-----------------------------------------------#
# metaQiitaWGS, metaQiitaRNA
sum(colnames(metaQiitaWGS) %in% colnames(metaQiitaRNA)) # 28
colnames(metaQiitaRNA)[!(colnames(metaQiitaRNA) %in% colnames(metaQiitaWGS))]
colnames(metaQiitaWGS)[!(colnames(metaQiitaWGS) %in% colnames(metaQiitaRNA))]
colnames(metaQiitaWGS)[which(colnames(metaQiitaWGS) == "id_cgc")] <- "cgc_id"
colnames(metaQiitaWGS)[which(colnames(metaQiitaWGS) == "filename_cgc")] <- "cgc_filename"
colnames(metaQiitaWGS)[which(colnames(metaQiitaWGS) == "sample_id_tcga")] <- "tcga_sample_id"
colnames(metaQiitaWGS)[which(colnames(metaQiitaWGS) == "case_uuid_cgc")] <- "cgc_case_uuid"
colnames(metaQiitaWGS)[which(colnames(metaQiitaWGS) == "platform_tcga")] <- "cgc_platform"
colnames(metaQiitaWGS)[which(colnames(metaQiitaWGS) == "gdc_uuid")] <- "gdc_file_uuid"
colnames(metaQiitaWGS)[which(colnames(metaQiitaWGS) == "sample_uuid_cgc")] <- "cgc_sample_uuid"
colnames(metaQiitaWGS)[which(colnames(metaQiitaWGS) == "aliquot_uuid_cgc")] <- "cgc_aliquot_uuid"
colnames(metaQiitaWGS)[which(colnames(metaQiitaWGS) == "aliquot_id_tcga")] <- "tcga_aliquot_id"
colnames(metaQiitaWGS)[which(colnames(metaQiitaWGS) == "case_id_tcga")] <- "tcga_case_id"
colnames(metaQiitaWGS)[which(colnames(metaQiitaWGS) == "base_name")] <- "cgc_base_name"
# Sanity check
sum(colnames(metaQiitaWGS) %in% colnames(metaQiitaRNA)) # 39
# Intersect overlapping metadata columns and rbind
intersectingMetadataColumns <- intersect(colnames(metaQiitaWGS),
colnames(metaQiitaRNA))
metaQiitaWGS_RNA_AllSeqPlatforms_Joined <- rbind(metaQiitaWGS[,intersectingMetadataColumns],
metaQiitaRNA[,intersectingMetadataColumns])
dim(metaQiitaWGS_RNA_AllSeqPlatforms_Joined) # 15512 39
metaQiitaWGS_RNA_AllSeqPlatforms_Joined %>% count(data_submitting_center_label)
missingSeqCenterAllSeqPlatforms <- rownames(metaQiitaWGS_RNA_AllSeqPlatforms_Joined)[which(is.na(metaQiitaWGS_RNA_AllSeqPlatforms_Joined$data_submitting_center_label))]
metaQiitaWGS_RNA_AllSeqPlatforms_Joined_Filt <- droplevels(metaQiitaWGS_RNA_AllSeqPlatforms_Joined[!(rownames(metaQiitaWGS_RNA_AllSeqPlatforms_Joined) %in% missingSeqCenterAllSeqPlatforms),])
dim(metaQiitaWGS_RNA_AllSeqPlatforms_Joined_Filt) # 15484 39
# save(metaQiitaWGS_RNA_AllSeqPlatforms_Joined,
# metaQiitaWGS_RNA_AllSeqPlatforms_Joined_Filt,
# file = "Interim_data/metaQiitaWGS_RNA_AllSeqPlatforms_Joined_25Mar22.RData")
#-----------------------------------------------#
# Combine WGS and RNA data - HiSeq only
#-----------------------------------------------#
dim(rep200Data_WGS_HiSeq_Fungi) # 4387 318
dim(rep200Data_RNA_HiSeq_Fungi) # 10701 319
dim(metaQiitaWGS_HiSeq_Filt) # 4387 39
dim(metaQiitaRNA_HiSeq_Filt) # 10701 41
sum(colnames(rep200Data_WGS_HiSeq_Fungi) %in% colnames(rep200Data_RNA_HiSeq_Fungi)) # 318
sum(colnames(metaQiitaWGS_HiSeq_Filt) %in% colnames(metaQiitaRNA_HiSeq_Filt)) # 28
colnames(metaQiitaRNA_HiSeq_Filt)[!(colnames(metaQiitaRNA_HiSeq_Filt) %in% colnames(metaQiitaWGS_HiSeq_Filt))]
colnames(metaQiitaWGS_HiSeq_Filt)[!(colnames(metaQiitaWGS_HiSeq_Filt) %in% colnames(metaQiitaRNA_HiSeq_Filt))]
colnames(metaQiitaWGS_HiSeq_Filt)[which(colnames(metaQiitaWGS_HiSeq_Filt) == "id_cgc")] <- "cgc_id"
colnames(metaQiitaWGS_HiSeq_Filt)[which(colnames(metaQiitaWGS_HiSeq_Filt) == "filename_cgc")] <- "cgc_filename"
colnames(metaQiitaWGS_HiSeq_Filt)[which(colnames(metaQiitaWGS_HiSeq_Filt) == "sample_id_tcga")] <- "tcga_sample_id"
colnames(metaQiitaWGS_HiSeq_Filt)[which(colnames(metaQiitaWGS_HiSeq_Filt) == "case_uuid_cgc")] <- "cgc_case_uuid"
colnames(metaQiitaWGS_HiSeq_Filt)[which(colnames(metaQiitaWGS_HiSeq_Filt) == "platform_tcga")] <- "cgc_platform"
colnames(metaQiitaWGS_HiSeq_Filt)[which(colnames(metaQiitaWGS_HiSeq_Filt) == "gdc_uuid")] <- "gdc_file_uuid"
colnames(metaQiitaWGS_HiSeq_Filt)[which(colnames(metaQiitaWGS_HiSeq_Filt) == "sample_uuid_cgc")] <- "cgc_sample_uuid"
colnames(metaQiitaWGS_HiSeq_Filt)[which(colnames(metaQiitaWGS_HiSeq_Filt) == "aliquot_uuid_cgc")] <- "cgc_aliquot_uuid"
colnames(metaQiitaWGS_HiSeq_Filt)[which(colnames(metaQiitaWGS_HiSeq_Filt) == "aliquot_id_tcga")] <- "tcga_aliquot_id"
colnames(metaQiitaWGS_HiSeq_Filt)[which(colnames(metaQiitaWGS_HiSeq_Filt) == "case_id_tcga")] <- "tcga_case_id"
colnames(metaQiitaWGS_HiSeq_Filt)[which(colnames(metaQiitaWGS_HiSeq_Filt) == "base_name")] <- "cgc_base_name"
# Sanity check
sum(colnames(metaQiitaWGS_HiSeq_Filt) %in% colnames(metaQiitaRNA_HiSeq_Filt)) # 39
# Intersect overlapping metadata columns and rbind
intersectingMetadataColumns <- intersect(colnames(metaQiitaWGS_HiSeq_Filt),
colnames(metaQiitaRNA_HiSeq_Filt))
metaQiitaWGS_RNA_HiSeq_Joined <- rbind(metaQiitaWGS_HiSeq_Filt[,intersectingMetadataColumns],
metaQiitaRNA_HiSeq_Filt[,intersectingMetadataColumns])
dim(metaQiitaWGS_RNA_HiSeq_Joined) # 15088 39
# Rbind fungi data
require(gtools)
colnames(rep200Data_RNA_HiSeq_Fungi)[!(colnames(rep200Data_RNA_HiSeq_Fungi) %in% colnames(rep200Data_WGS_HiSeq_Fungi))] # G000277815
rep200Data_WGS_RNA_HiSeq_Fungi <- smartbind(cbind(rep200Data_WGS_HiSeq_Fungi, G000277815=0),
rep200Data_RNA_HiSeq_Fungi)
rownames(rep200Data_WGS_RNA_HiSeq_Fungi) <- c(rownames(rep200Data_WGS_HiSeq_Fungi),
rownames(rep200Data_RNA_HiSeq_Fungi))
dim(rep200Data_WGS_RNA_HiSeq_Fungi) # 15088 319
rep200Data_WGS_HiSeq_Fungi[1:3,1:3]
# Subset WGS and RNA data and save --> to be used for alpha rarefaction
metaQiitaWGS_RNA_HiSeq_Joined_WGS <- metaQiitaWGS_RNA_HiSeq_Joined %>% filter(experimental_strategy == "WGS") %>% droplevels()
metaQiitaWGS_RNA_HiSeq_Joined_RNA <- metaQiitaWGS_RNA_HiSeq_Joined %>% filter(experimental_strategy == "RNA-Seq") %>% droplevels()
rep200Data_WGS_RNA_HiSeq_Fungi_WGS <- rep200Data_WGS_RNA_HiSeq_Fungi[rownames(metaQiitaWGS_RNA_HiSeq_Joined_WGS),]
rep200Data_WGS_RNA_HiSeq_Fungi_RNA <- rep200Data_WGS_RNA_HiSeq_Fungi[rownames(metaQiitaWGS_RNA_HiSeq_Joined_RNA),]
# save(metaQiitaWGS_RNA_HiSeq_Joined,
# rep200Data_WGS_RNA_HiSeq_Fungi,
# metaQiitaWGS_RNA_HiSeq_Joined_WGS,
# rep200Data_WGS_RNA_HiSeq_Fungi_WGS,
# metaQiitaWGS_RNA_HiSeq_Joined_RNA,
# rep200Data_WGS_RNA_HiSeq_Fungi_RNA,
# file = "Interim_data/data_for_alpha_rarefaction_25Mar22.RData")
metaQiitaWGS_RNA_HiSeq_Joined %>% count(data_submitting_center_label)
missingSeqCenter <- rownames(metaQiitaWGS_RNA_HiSeq_Joined)[which(is.na(metaQiitaWGS_RNA_HiSeq_Joined$data_submitting_center_label))]
smallHopkins <- rownames(metaQiitaWGS_RNA_HiSeq_Joined)[metaQiitaWGS_RNA_HiSeq_Joined$data_submitting_center_label == "Johns Hopkins / University of Southern California"]
samples2Remove <- c(missingSeqCenter, na.omit(smallHopkins))
metaQiitaWGS_RNA_HiSeq_Filt <- droplevels(metaQiitaWGS_RNA_HiSeq_Joined[!(rownames(metaQiitaWGS_RNA_HiSeq_Joined) %in% samples2Remove),])
dim(metaQiitaWGS_RNA_HiSeq_Filt) # 15059 39
#---------------------------------------------#
# Using plate number for batch to decontaminate
#---------------------------------------------#
# Function for extracting last n characters from R string
# URL: https://stackoverflow.com/questions/7963898/extracting-the-last-n-characters-from-a-string-in-r
substrRight <- function(x, n){substr(x, nchar(x)-n+1, nchar(x))}
# NB: metaQiitaWGS_RNA_HiSeq_Filt IS A METADATA FILE OF TCGA SAMPLES CONTAINING ALIQUOT IDS FOR ALL SAMPLES
# IN A COLUMN CALLED "tcga_aliquot_id"
tmp <- as.character(metaQiitaWGS_RNA_HiSeq_Filt$tcga_aliquot_id)
metaQiitaWGS_RNA_HiSeq_Filt$PlateCenter <- factor(substrRight(tmp, 7))
# NB: SINCE DECONTAM ESSENTIALLY PERFORMS LINEAR REGRESSION BETWEEN READ FRACTIONS AND
# ANALYTE CONCENTRATIONS, AT LEAST 10 SAMPLES ARE REQUIRED PER PLATE-CENTER COMBINATION
# TO BE PROCESSED FOR IDENTIFYING PUTATIVE CONTAMINANTS. NOTE ALSO THAT ANY CONTAMINANT
# IDENTIFIED IN ANY ONE PLATE-CENTER BATCH WILL BE REMOVED FROM THE WHOLE DATASET
booleanPlateCenter <- as.logical(table(metaQiitaWGS_RNA_HiSeq_Filt$PlateCenter)>=10)
sufficientPlateCenter <- names(table(metaQiitaWGS_RNA_HiSeq_Filt$PlateCenter))[booleanPlateCenter]
length(sufficientPlateCenter) # 329
metaQiitaWGS_RNA_HiSeq_Filt$PlateCenterFlag <- (metaQiitaWGS_RNA_HiSeq_Filt$PlateCenter %in% sufficientPlateCenter)
metaQiitaWGS_RNA_HiSeq_Filt_PlateCenterSubset <- droplevels(metaQiitaWGS_RNA_HiSeq_Filt[metaQiitaWGS_RNA_HiSeq_Filt$PlateCenterFlag &
!is.na(metaQiitaWGS_RNA_HiSeq_Filt$aliquot_concentration),])
dim(metaQiitaWGS_RNA_HiSeq_Filt_PlateCenterSubset) # 14374 41
# NB: rep200Data_WGS_RNA_HiSeq_Fungi_PlateCenterSubset CONTAINS RAW TCGA FUNGI DATA FROM rep200
rep200Data_WGS_RNA_HiSeq_Fungi_PlateCenterSubset <- rep200Data_WGS_RNA_HiSeq_Fungi[rownames(metaQiitaWGS_RNA_HiSeq_Filt_PlateCenterSubset),]
# Decontam
require(decontam)
countDataPlateCenter <- rep200Data_WGS_RNA_HiSeq_Fungi_PlateCenterSubset
countMetadataPlateCenter <- metaQiitaWGS_RNA_HiSeq_Filt_PlateCenterSubset
# Remove zero sum samples (25 Mar 22) by identifying and removing indices
zeroSumIdxPlateCenter <- unname(which(rowSums(countDataPlateCenter)==0))
countDataPlateCenterNonzero <- countDataPlateCenter[-zeroSumIdxPlateCenter,]
countMetadataPlateCenterNonzero <- countMetadataPlateCenter[-zeroSumIdxPlateCenter,]
dim(countDataPlateCenterNonzero) # 13932 319
dim(countMetadataPlateCenterNonzero) # 13932 41
contamdf.freq.fungi.plateCenter <- isContaminant(seqtab = as.matrix(countDataPlateCenterNonzero),
conc = countMetadataPlateCenterNonzero$aliquot_concentration,
method = "frequency",
batch = countMetadataPlateCenterNonzero$PlateCenter,
threshold = 0.1) # DEFAULT VALUE IS 0.1
save(contamdf.freq.fungi.plateCenter, file = "Interim_data/contamdf.freq.fungi.plateCenter_25Mar22.RData")
# load("Interim_data/contamdf.freq.fungi.plateCenter_25Mar22.RData")
table(contamdf.freq.fungi.plateCenter$contaminant) # 57 TRUE | 262 FALSE
hist(contamdf.freq.fungi.plateCenter$p)
contaminants <- rownames(contamdf.freq.fungi.plateCenter)[contamdf.freq.fungi.plateCenter$contaminant]
contaminantsFungiPlateCenterTCGA <- rep200TaxSplit_Fungi[contaminants,]
# save(contaminantsFungiPlateCenterTCGA, file = "Interim_data/contaminantsFungiPlateCenterTCGA_25Mar22.RData")
notContamSumFreq <- colSums(as.matrix(rep200Data_WGS_RNA_HiSeq_Fungi)[,!contamdf.freq.fungi.plateCenter$contaminant])
contamSumFreq <- colSums(as.matrix(rep200Data_WGS_RNA_HiSeq_Fungi)[,contamdf.freq.fungi.plateCenter$contaminant])
sum(contamSumFreq)/sum(colSums(as.matrix(rep200Data_WGS_RNA_HiSeq_Fungi))) #--> 0.008572445 (25 Mar 22)
#------------------------Version 2 of decontamination------------------------#
## Save data for literature searching
# Added later. Goal is to cross-examine decontam results with
# biological plausibility (from the literature) and WIS results
load("Interim_data/shared_fungi_features_at_each_taxa_level_13Sep21.RData", verbose=TRUE) ## Load shared features with Weizmann
rep200TaxSplit_Fungi_Paired_to_Weizmann <- read.csv("Supporting_data/rep200TaxSplit_Fungi_Paired_To_Weizmann_Final.csv", stringsAsFactors = FALSE, row.names = 1)
decontamResults <- contamdf.freq.fungi.plateCenter
decontamResults$species <- rep200TaxSplit_Fungi_Paired_to_Weizmann[rownames(decontamResults),"species"]
decontamResults$sharedWIS <- ifelse(decontamResults$species %in% sharedSpecies, yes = "YES", no = "NO")
# decontamResults %>% write.csv(file = "Interim_data/contaminantsFungiPlateCenterTCGA_updated_annotations_25Mar22.csv")
## Load data after literature searching is complete
decontamResultsV2 <- read.csv("Supporting_data/mycobiome_contaminant_analyses_updated_annotations_25Mar22.csv", row.names = 1, stringsAsFactors = FALSE)
# Extract PMIDs
require(stringr)
uniquePMIDs <- na.omit(unique(str_extract(string = decontamResultsV2$comments_and_literature, "[0-9]{7,8}")))
length(uniquePMIDs)
# Process rest of decontamV2
decontamResultsV2 <- decontamResultsV2[,!(colnames(decontamResultsV2) == "comments_and_literature")]
decontamResultsV2$decision <- ifelse(decontamResultsV2$shared_with_WIS == "YES" |
decontamResultsV2$in_hmp_gut_mycobiome_metagenomic_data %in% c("YES","YES*") |
decontamResultsV2$known_human_association_literature == "YES" |
(decontamResultsV2$known_human_association_literature == "UNKNOWN" &
decontamResultsV2$decontam_predicted_contaminant == "FALSE"),
yes = "KEEP", no = "DISCARD")
table(decontamResultsV2$decision) # KEEP 224 | DISCARD 95
contaminantsV2 <- rownames(decontamResultsV2)[which(decontamResultsV2$decision == "DISCARD")]
decontamResultsV2$reason <- decontamResultsV2$known_human_association_literature
decontamResultsV2$reason[decontamResultsV2$known_human_association_literature == "YES"] <- "Known human association"
decontamResultsV2$reason[decontamResultsV2$known_human_association_literature == "NO"] <- "No known human association"
decontamResultsV2$reason[decontamResultsV2$known_human_association_literature == "UNKNOWN" & decontamResultsV2$decontam_predicted_contaminant == "FALSE"] <- "Unknown human association but\nnot predicted contaminant"
decontamResultsV2$reason[decontamResultsV2$known_human_association_literature == "UNKNOWN" & decontamResultsV2$decontam_predicted_contaminant == "TRUE"] <- "Unknown human association and\npredicted contaminant"
decontamResultsV2$reason[decontamResultsV2$in_hmp_gut_mycobiome_metagenomic_data %in% c("YES","YES*")] <- "In HMP gut mycobiome data"
decontamResultsV2$reason[decontamResultsV2$shared_with_WIS == "YES"] <- "Shared with WIS"
decontamResultsV2$reason <- factor(decontamResultsV2$reason, levels = c("Shared with WIS","In HMP gut mycobiome data","Known human association",
"Unknown human association but\nnot predicted contaminant",
"Unknown human association and\npredicted contaminant",
"No known human association"))
save(decontamResultsV2, contaminantsV2, file = "Interim_data/decontamResultsV2_25Mar22.RData")
contaminantsV2boolean <- ifelse(colnames(rep200Data_WGS_RNA_HiSeq_Fungi) %in% contaminantsV2,
yes = TRUE, no = FALSE)
notContamSumFreq <- colSums(as.matrix(rep200Data_WGS_RNA_HiSeq_Fungi)[,!contaminantsV2boolean])
contamSumFreq <- colSums(as.matrix(rep200Data_WGS_RNA_HiSeq_Fungi)[,contaminantsV2boolean])
sum(contamSumFreq)/sum(colSums(as.matrix(rep200Data_WGS_RNA_HiSeq_Fungi))) #--> 0.02139181 (25 Mar 22)
#-------------------------------------------------------------#
rep200Data_WGS_RNA_HiSeq_Fungi_DecontamV2 <- rep200Data_WGS_RNA_HiSeq_Fungi[,!(colnames(rep200Data_WGS_RNA_HiSeq_Fungi) %in% contaminantsV2)]
dim(rep200Data_WGS_RNA_HiSeq_Fungi_DecontamV2) # 15088 224
sum(rowSums(rep200Data_WGS_RNA_HiSeq_Fungi_DecontamV2) == 0) # 564
emptySamplesAfterDecontamV2 <- names(which(rowSums(rep200Data_WGS_RNA_HiSeq_Fungi_DecontamV2) == 0))
samples2RemoveWithDecontamV2 <- c(emptySamplesAfterDecontamV2, missingSeqCenter, na.omit(smallHopkins))
rep200Data_WGS_RNA_HiSeq_Fungi_DecontamV2_Nonzero <- rep200Data_WGS_RNA_HiSeq_Fungi_DecontamV2[!(rownames(rep200Data_WGS_RNA_HiSeq_Fungi_DecontamV2) %in% samples2RemoveWithDecontamV2),]
metaQiitaWGS_RNA_HiSeq_Filt_Nonzero_DecontamV2 <- droplevels(metaQiitaWGS_RNA_HiSeq_Filt[!(rownames(metaQiitaWGS_RNA_HiSeq_Filt) %in% samples2RemoveWithDecontamV2),])
dim(rep200Data_WGS_RNA_HiSeq_Fungi_DecontamV2_Nonzero) # 14495 224
dim(metaQiitaWGS_RNA_HiSeq_Filt_Nonzero_DecontamV2) # 14495 41
## Convert strings to factors
metaQiitaWGS_RNA_HiSeq_Filt_Nonzero_DecontamV2$sample_type <- factor(metaQiitaWGS_RNA_HiSeq_Filt_Nonzero_DecontamV2$sample_type)
metaQiitaWGS_RNA_HiSeq_Filt_Nonzero_DecontamV2$disease_type <- factor(metaQiitaWGS_RNA_HiSeq_Filt_Nonzero_DecontamV2$disease_type)
metaQiitaWGS_RNA_HiSeq_Filt_Nonzero_DecontamV2$experimental_strategy <- factor(metaQiitaWGS_RNA_HiSeq_Filt_Nonzero_DecontamV2$experimental_strategy)
metaQiitaWGS_RNA_HiSeq_Filt_Nonzero_DecontamV2$data_submitting_center_label <- factor(metaQiitaWGS_RNA_HiSeq_Filt_Nonzero_DecontamV2$data_submitting_center_label)
metaQiitaCombined_Nonzero_DecontamV2 <- metaQiitaWGS_RNA_HiSeq_Filt_Nonzero_DecontamV2
save(rep200Data_WGS_RNA_HiSeq_Fungi_DecontamV2_Nonzero,
metaQiitaCombined_Nonzero_DecontamV2,
file = "Interim_data/fungi_data_WGS_RNA_HiSeq_DecontamV2_25Mar22.RData")
#------------------------------------------------------------------#
# WGS vs. RNA-Seq norm read counts
#------------------------------------------------------------------#
metaQiitaCombined_Nonzero_DecontamV2_DNAonly <- metaQiitaCombined_Nonzero_DecontamV2 %>% filter(experimental_strategy == "WGS") %>% droplevels()
metaQiitaCombined_Nonzero_DecontamV2_RNAonly <- metaQiitaCombined_Nonzero_DecontamV2 %>% filter(experimental_strategy == "RNA-Seq") %>% droplevels()
wgsData <- rep200Data_WGS_RNA_HiSeq_Fungi_DecontamV2_Nonzero[rownames(metaQiitaCombined_Nonzero_DecontamV2_DNAonly),]
rnaData <- rep200Data_WGS_RNA_HiSeq_Fungi_DecontamV2_Nonzero[rownames(metaQiitaCombined_Nonzero_DecontamV2_RNAonly),]
wgsSampleCounts <- log10(unname(rowSums(wgsData)))
rnaSampleCounts <- log10(unname(rowSums(rnaData)))
summary(wgsSampleCounts)
summary(rnaSampleCounts)
combinedSampleCounts <- data.frame(sample_counts = c(wgsSampleCounts,rnaSampleCounts),
data_type = c(rep("WGS",length(wgsSampleCounts)), rep("RNA-Seq",length(rnaSampleCounts))))
require(EnvStats)
combinedSampleCounts %>%
ggviolin(x = "data_type",
y = "sample_counts",
fill = "data_type",
palette = "nejm",
legend = "none",
draw_quantiles = c(0.25,0.50,0.75),
xlab = "Experimental strategy",
ylab = "log10(decontaminated sample fungi read counts)",
add = "mean",
add.params = list(color="white",size=1)) +
stat_compare_means(label.x.npc = 0.1, label.y = 7) +
stat_n_text()
ggsave(filename = "Figures/Supplementary_Figures/tcga_read_count_wgs_vs_rna_decontamV2.pdf",
dpi = "retina",
height = 5,
width = 3,
units = "in")
#-----------------------------------------------#
# Voom-SNM
#-----------------------------------------------#
require(limma)
require(edgeR)
require(dplyr)
require(snm)
qcMetadata <- metaQiitaCombined_Nonzero_DecontamV2
qcData <- rep200Data_WGS_RNA_HiSeq_Fungi_DecontamV2_Nonzero
qcMetadata %>% count(data_submitting_center_label)
# Set up design matrix
covDesignNorm <- model.matrix(~0 + sample_type +
data_submitting_center_label +
experimental_strategy,
data = qcMetadata)
# Check row dimensions
dim(covDesignNorm)[1] == dim(qcData)[1]
print(colnames(covDesignNorm))
colnames(covDesignNorm) <- gsub('([[:punct:]])|\\s+','',colnames(covDesignNorm))
print(colnames(covDesignNorm))
# Set up counts matrix
counts <- t(qcData) # DGEList object from a table of counts (rows=features, columns=samples)
# Normalize using edgeR and then plug into voom
dge <- DGEList(counts = counts)
vdge_data <- voom(dge, design = covDesignNorm, plot = TRUE, save.plot = TRUE,
normalize.method="quantile")
vdge_dataE_DecontamV2 <- t(vdge_data$E)
# Apply
bio.var <- model.matrix(~sample_type,
data=qcMetadata)
adj.var <- model.matrix(~data_submitting_center_label +
experimental_strategy,
data=qcMetadata)
colnames(bio.var) <- gsub('([[:punct:]])|\\s+','',colnames(bio.var))
colnames(adj.var) <- gsub('([[:punct:]])|\\s+','',colnames(adj.var))
print(dim(adj.var))
print(dim(bio.var))
print(dim(t(vdge_data$E)))
print(dim(covDesignNorm))
snmDataObjOnly <- snm(raw.dat = vdge_data$E,
bio.var = bio.var,
adj.var = adj.var,
rm.adj=TRUE,
verbose = TRUE,
diagnose = TRUE)
snmDataOGUFungiDecontamV2 <- t(snmDataObjOnly$norm.dat)
save(snmDataOGUFungiDecontamV2,
vdge_dataE_DecontamV2,
rep200Data_WGS_RNA_HiSeq_Fungi_DecontamV2_Nonzero,
metaQiitaCombined_Nonzero_DecontamV2,
file = "Interim_data/snmDataFungi_DecontamV2_25Mar22.RData")
#------------------------------------------------------------------#
# PVCA
#------------------------------------------------------------------#
# The following scripts were run on a compute cluster to generate results:
# "Supplementary_scripts/S01B-Run-pvca-fungi.R"
# "Supplementary_scripts/S02-pvca-function.R"
# load("Interim_data/pvca_fungi_results_raw_Voom_VSNM_13Sep21.RData")
#
# pvcaRawRound <- round(pvcaRaw,3)
# pvcaVoomRound <- round(pvcaVoom,3)
# pvcaVSNMRound <- round(pvcaVSNM,3)
#
# pvcaRes <- data.frame('Sample Type' = c(pvcaRawRound[1], pvcaVoomRound[1], pvcaVSNMRound[1]),
# 'Disease Type' = c(pvcaRawRound[2], pvcaVoomRound[2], pvcaVSNMRound[2]),
# 'Sequencing Center' = c(pvcaRawRound[3], pvcaVoomRound[3], pvcaVSNMRound[3]),
# 'Experimental Strategy' = c(pvcaRawRound[4], pvcaVoomRound[4], pvcaVSNMRound[4]),
# 'Residual\n(not explained by\ntechnical variation)' = c(pvcaRawRound[5], pvcaVoomRound[5], pvcaVSNMRound[5]),
# data_type = factor(c("Raw count data","Voom Normalized Data","Voom Normalized & SNM Corrected Data"),
# levels = c("Raw count data","Voom Normalized Data","Voom Normalized & SNM Corrected Data")),
# check.names = FALSE)
#
# pvcaRes.melted <- reshape2::melt(pvcaRes, id.vars = "data_type")
# pvcaRes.melted %>%
# ggbarplot(x = "variable",
# y = "value",
# fill = "data_type",
# palette = "nejm",
# legend = "top",
# ylim = c(0,1),
# xlab = "Biological Effects & Technical Effects",
# ylab = "Weighted average proportion variance",
# label = TRUE,
# position = position_dodge(0.9)) +
# labs(fill = "Data type") +
# ggsave("Figures/Supplementary_Figures/pvca_plot_OGUs_13Sep21.jpeg",
# dpi = "retina",
# width = 12,
# height = 3,
# units = "in")