-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy path04-assoc_spirometry.R
executable file
·311 lines (289 loc) · 12.9 KB
/
04-assoc_spirometry.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
# Association analysis of inverse rank normal transformed (IRNT)
# spirometry measured in GOLD2-4 spirometry-defined COPD patients in the UKBB around SLC26A9 locus
library(data.table)
library(rbgen)
library(RNOmni)
library(bigmemory)
get_irnt <- function(vec) {
narows <- which(is.na(vec))
if(length(narows)>0) vec <- vec[-narows]
vecRankNorm <- rep(NA, length(vec))
vecRankNorm <- rankNorm(vec)
if(length(narows)>0) vecRankNorm[-narows] <- rankNorm(vec)
return(vecRankNorm)
}
calculate_dr2 <- function(y1,y2) {
if(length(y1)!=length(y2)) stop("vector lengths differ")
u <- y1 + 2*y2
z <- u
w <- y1 + 4*y2
n <- length(y1)
numerator <- ( sum(z*u) - (1/n)*(sum(u)*sum(z)) )^2
denominator <- (sum(w) - (1/n)*(sum(u))^2) * (sum(z^2) - (1/n)*(sum(z))^2)
r2 <- numerator / denominator
return(r2)
}
convert_to_eid <- function(anonymous_sample_name) {
if(length(anonymous_sample_name)!=1 & class(anonymous_sample_name)!="character") {
stop("input must be of length 1 and character type")
}
return(pheno$eid[which(pheno$samplename %in% anonymous_sample_name)])
}
get_dosagemat <- function(bgen_filename, ranges, samples, chunksize, fileout) {
rangestr <- with(ranges, paste0(as.character(chromosome),":",as.character(start),"-",as.character(end)))
print(paste0("Getting number of SNPs in ", rangestr))
tmp <- bgen.load(bgen_filename, ranges, samples = samples[1:10])
mapdf <- data.table(tmp$variants)
fwrite(mapdf, paste0(fileout,".map"), sep="\t", quote=F)
numsnps <- dim(tmp$data)[1]
numsamples <- length(samples)
print(paste0("Initializing big matrix with dimension: ",numsamples," x ",numsnps))
dat <- big.matrix(nrow=numsamples, ncol=numsnps, type="double", init=-9.0)
#dat <- data.table(matrix(rep(-9.0,numsnps*numsamples), nrow=numsnps, ncol=numsamples))
samplecounter <- 1
while(samplecounter <= numsamples) {
starti <- samplecounter
endi <- starti + chunksize - 1
if(endi>numsamples) endi <- numsamples
ivcfmat <- matrix(rep("",(endi-starti+1)*numsnps), nrow=endi-starti+1, ncol=numsnps) # inverted vcf data contents (sample x snp)
print(paste0("Getting samples ",starti," to ",endi, " out of ", numsamples))
currdata <- bgen.load(bgen_filename, ranges, samples = samples[starti:endi])
currsamplenames <- dimnames(currdata$data)[[2]]
currsamplenames <- unname(sapply(currsamplenames, convert_to_eid))
print(paste0("Writing sample eid's to: ", paste0(fileout,".sample")))
fwrite(data.table(eid=currsamplenames), paste0(fileout,".sample"), sep="\t", quote=F, append=TRUE)
for(j in 1:numsnps){
if(j==1 | (j %% 100)==0) {
print(paste0("SNP ",j," of ",numsnps,". "
,format(round((j/numsnps)*100,2),nsmall=2)
,"% done"))
}
snpi <- currdata$data[j,,]
dosage <- round(snpi[,"g=1"] + 2*snpi[,"g=2"], 3)
dat[starti:endi,j] <- dosage
maxGP <- unname(unlist(apply(snpi, 1, function(x) which(x==max(x))[1]-1)))
names(maxGP) <- rownames(snpi)
GT <- ifelse(maxGP==0, "0/0", ifelse(maxGP==1, "0/1", ifelse(maxGP==2, "1/1", NA)))
DS <- dosage
GP <- apply(snpi, 1, function(x) paste0(round(x,3),collapse=","))
ivcfmat[,j] <- paste0(GT,":",DS,":",GP)
#currsnpnames <- dimnames(currdata$data)[[1]]
#colnames(ivcfmat) <- currsnpnames
#rownames(ivcfmat) <- currsamplenames
}
print(paste0("Writing imputation data to: ", paste0(fileout,".dat")))
fwrite(ivcfmat, paste0(fileout,".dat"), sep="\t", quote=F, append=TRUE)
samplecounter <- endi+1
}
return(dat)
}
fileout <- "data/intermediate_files/ukbb_imputation_slc26a9/ukbb_spiroqc_v3"
outdir <- "data/clean/assoc/"
print("Loading spirometry phenotypes")
pheno <- fread("data/clean/ukbb_spiro_and_geno_qc_v2.csv") # 264,273
pheno[,ratio := fev1/fvc]
pheno[,eid := as.character(eid)]
print("Loading PCA")
pca <- fread("data/intermediate_files/pca/18-ukbb_ukbbspiro_flashpca2_eigenvectors.txt", header=F, stringsAsFactors = F) # 264,273
pca <- pca[,c("V1","V2","V3","V4"),with=F]
pca[,V1 := gsub("(.*):(.*)","\\2",V1)]
colnames(pca) <- c("eid","PC1","PC2","PC3")
covar <- pheno[,c("eid","sex","age", "everSmoked")]
covar <- cbind(covar, age2=covar$age^2)
covar[,eid := as.character(eid)]
print("Merging phenotypes and PCA")
pheno <- merge(pheno,pca,by="eid") # 264,273
pheno[,ratio.irnt := get_irnt(ratio)]
pheno[,fev1pp.irnt := get_irnt(fev1pp)]
pheno[,pef.irnt := get_irnt(pef)]
print(paste0("Loading imputation data for ",nrow(pheno)," samples"))
#data = bgen.load( "/hpf/largeprojects/struglis/datasets/uk_biobank_40946/imputation/ukb_imp_chr1_v3.bgen", ranges, samples = pheno$samplename) # too big to load
bgen_filename <- "/hpf/largeprojects/struglis/datasets/uk_biobank_40946/imputation/ukb_imp_chr1_v3.bgen"
ranges = data.frame(
chromosome = "01",
start = 205780000,
end = 205940000
)
samples <- pheno$samplename
# If running for the first time, uncomment next 3 lines
# and comment the lines loading them (subsequent 3 lines):
#chunksize <- 70000
#dosage <- get_dosagemat(bgen_filename, ranges, samples, chunksize, fileout)
#write.big.matrix(dosage, paste0(fileout,"_dosage.bigmat"), sep="\t")
# If the dosage matrix has already been created before,
# simply load the files in the next 3 lines:
dosage <- read.big.matrix(paste0(fileout,"_dosage.bigmat"), sep="\t", type="double")
map <- fread(paste0(fileout,".map")) # generated above with get_dosagemat function
sample <- fread(paste0(fileout,".sample")) # generated above with get_dosagemat function
# Association with all 264,273 individuals
print(paste0("Starting association analysis for ",nrow(pheno)," samples and ",nrow(map)," variants"))
phenocols <- c("ratio.irnt"
,"fev1pp.irnt"
,"hasCOPD"
,"pef.irnt")
results <- NULL
for(j in 1:length(phenocols)) {
phenoj <- pheno[,c("eid", phenocols[j]),with=F]
phenoname <- phenocols[j]
print(paste0("Running association analysis for ", phenoname))
phenoj_assoc <- NULL
# for each snp
for(i in 1:ncol(dosage)) {
if(i==1 | (i %% 100)==0) {
print(paste0("SNP ",i," of ",ncol(dosage),". "
,format(round((i/ncol(dosage))*100,2),nsmall=2), "% done"))
}
dosagei <- dosage[,i]
af <- sum(dosagei,na.rm=T) / (2*length(dosagei))
maf <- af
if(maf>0.5) maf <- 1-maf
if(maf<(10/length(dosagei))) next
snpi <- data.table(cbind(sample, dosagei))
snpi[,eid := as.character(eid)]
x <- merge(phenoj, covar, by="eid")
x <- merge(x, pca, by="eid")
x <- merge(x, snpi[,c("eid","dosagei")], by="eid")
#x <- x[,-c("eid"),with=F]
x <- na.omit(x)
x[,dosagei := as.numeric(dosagei)]
x[,sex := as.factor(sex)]
realmaf <- sum(x$dosagei,na.rm=T)/(2*nrow(x))
if(realmaf>0.5) maf <- 1-maf
if(realmaf<(10/nrow(x))) next
pcol <- 'Pr(>|t|)'
if(phenoname %in% "hasCOPD") {
model <- glm(formula(paste0(phenoname," ~ dosagei + PC1 + PC2 + PC3 + sex + age + age2 + sex:age + sex:age2 + everSmoked")), family="binomial", data=x)
pcol <- 'Pr(>|z|)'
} else {
model <- lm(formula(paste0(phenoname," ~ dosagei + PC1 + PC2 + PC3 + sex + age + age2 + sex:age + sex:age2 + everSmoked")), data=x)
}
snp <- map[i,]
coefmat <- coef(summary(model))
assoc <- data.frame(chrom=as.integer(snp$chromosome)
,pos=snp$position
,rsid=as.character(snp$rsid)
,allele0=snp$allele0
,allele1=snp$allele1
,beta=coefmat['dosagei','Estimate']
,std.err=coefmat['dosagei','Std. Error']
,P=coefmat['dosagei',pcol]
,N=nrow(x)
,allele_freq=af
,phenotype=phenoname
)
phenoj_assoc <- rbind(phenoj_assoc, assoc)
results <- rbind(results, assoc)
}
outname <- paste0(outdir,phenoname,".assoc_v3.csv")
print(paste0("Saving results to: ", outname))
fwrite(phenoj_assoc, outname, quote=F, row.names=F, col.names=T)
}
outname <- paste0(outdir,"spirometry_assoc_ukbb_26a9_v3.csv")
print(paste0("Saving all spirometry association results to: ", outname))
fwrite(results, outname, quote=F, row.names=F, col.names=T)
#
#
# # Association with GOLD2-4 COPD individuals
# print("")
# print("Starting association for GOLD2-4 COPD subset")
# pheno <- pheno[,-c("ratio.irnt","fev1pp.irnt","pef.irnt","PC1","PC2","PC3"),with=F]
# pheno <- pheno[hasCOPD==TRUE]
#
# print("Loading PCA")
# pca <- fread("data/intermediate_files/pca/15-ukbb_copd_flashpca2_eigenvectors.txt", header=F, stringsAsFactors = F)
# pca <- pca[,c("V1","V2","V3","V4"),with=F]
# pca[,V1 := gsub("(.*):(.*)","\\2",V1)]
# colnames(pca) <- c("eid","PC1","PC2","PC3")
#
# print("Merging phenotypes with PCA")
# pheno <- merge(pheno,pca,by="eid")
# pheno[,ratio.irnt := get_irnt(ratio)]
# pheno[,fev1pp.irnt := get_irnt(fev1pp)]
# pheno[,pef.irnt := get_irnt(pef)]
# ranges = data.frame(
# chromosome = "01",
# start = 205780000,
# end = 205940000
# )
# print(paste0("Loading imputation data for ",nrow(pheno)," samples"))
# data = bgen.load( "/hpf/largeprojects/struglis/datasets/uk_biobank_40946/imputation/ukb_imp_chr1_v3.bgen", ranges, samples = pheno$samplename)
#
#
# covar <- pheno[,c("eid","sex","age", "everSmoked")]
# covar <- cbind(covar, age2=covar$age^2)
#
#
# # Dosage sample file:
# print("Loading imputation sample file")
# samplefile <- fread("/hpf/largeprojects/struglis/datasets/uk_biobank_40946/imputation/sample_files/ukb40946_imp_chr1_v3_s487324.sample")
# samplefile <- samplefile[-1,] # 487,409
# samplefile$index <- 1:nrow(samplefile)
# samplefile$samplename <- paste0("(anonymous_sample_", samplefile$index, ")")
# samplefile <- samplefile[,-c("missing","sex","ID_2")]
# setnames(samplefile, "ID_1","eid")
# #i <- which(samplefile$eid %in% pheno$eid) # 22,176
# #copd_samples <- paste0("(anonymous_sample_", i, ")")
#
#
# print(paste0("Starting association analysis for ",nrow(pheno)," samples and ",nrow(data$variants)," variants"))
# phenocols <- c("ratio.irnt","fev1pp.irnt", "pef.irnt")
# results <- NULL
# for(j in 1:length(phenocols)) {
# phenoj <- pheno[,c("eid", phenocols[j]),with=F]
# phenoname <- phenocols[j]
# print(paste0("Running association analysis for ", phenoname))
# phenoj_assoc <- NULL
# # for each snp
# for(i in 1:nrow(data$variants)) {
# if(i==1 | (i %% 100)==0) {
# print(paste0("SNP ",i," of ",nrow(data$variants),". "
# ,format(round((i/nrow(data$variants))*100,2),nsmall=2), "% done"))
# }
# snpi <- data$data[i,,]
# dosage <- snpi[,"g=1"] + 2*snpi[,"g=2"]
# af <- sum(dosage,na.rm=T) / (2*length(dosage))
# maf <- af
# if(maf>0.5) maf <- 1-maf
# if(maf<(10/length(dosage))) next
# snpi <- cbind(snpi,dosage)
# sample_indices <- as.integer(gsub("\\(anonymous_sample_(\\d+)\\)","\\1",names(dosage)))
# sample_ids <- as.character(samplefile$eid[sample_indices])
# snpi <- data.table(cbind(snpi, sample_ids))
# covar[,eid := as.character(eid)]
# x <- merge(phenoj, covar, by="eid")
# x <- merge(x, pca, by="eid")
# x <- merge(x, snpi[,c("sample_ids","dosage")], by.x="eid", by.y="sample_ids")
# x <- x[,-c("eid"),with=F]
# x <- na.omit(x)
# x[,dosage := as.numeric(dosage)]
# x[,sex := as.factor(sex)]
# realmaf <- sum(x$dosage,na.rm=T)/(2*nrow(x))
# if(realmaf>0.5) maf <- 1-maf
# if(realmaf<(10/nrow(x))) next
# model <- glm(formula(paste0(phenoname,"~ dosage + PC1 + PC2 + PC3 + sex + age + age2 + sex:age + sex:age2 + everSmoked")), data=x)
# snp <- data$variants[i,]
# coefmat <- coef(summary(model))
# assoc <- data.frame(chrom=as.integer(snp$chromosome)
# ,pos=snp$position
# ,rsid=as.character(snp$rsid)
# ,REF=snp$allele0
# ,ALT=snp$allele1
# ,beta=coefmat['dosage','Estimate']
# ,std.err=coefmat['dosage','Std. Error']
# ,P=coefmat['dosage','Pr(>|t|)']
# ,N=nrow(x)
# ,allele_freq=af
# ,phenotype=phenoname
# )
# phenoj_assoc <- rbind(phenoj_assoc, assoc)
# results <- rbind(results, assoc)
# }
# outname <- paste0(outdir,phenoname,".assoc_copd_only_v3.csv")
# print(paste0("Saving results to: ", outname))
# fwrite(phenoj_assoc, outname, quote=F, row.names=F, col.names=T)
# }
# outname <- paste0(outdir,"spirometry_assoc_ukbb_copd_only_26a9_v3.csv")
# print(paste0("Saving all results of COPD association to: ", outname))
# fwrite(results, outname, quote=F, row.names=F, col.names=T)
#
#