-
Notifications
You must be signed in to change notification settings - Fork 2
/
genesis_nullmodel.R
267 lines (214 loc) · 8.39 KB
/
genesis_nullmodel.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
#== Args
args<-commandArgs(TRUE)
#===mandatory parameters
outcome.name <- args[1]
outcome.type <- args[2] # Continuous or Dichotomous
covariate.string <- args[3]
pheno.file <- args[4]
genotype.file <- args[5]
results.file <- args[6]
#==optional parameters
kinship.matrix <- args[7]
pheno.id <- args[8]
# added these to JSON
test.stat <- args[9] # Score, Wald, Firth
conditional <- args[10] # 1:237733935:G:A
het_varsIn <- args[11]
# GLOBAL VARIABLES
collapsing.tests <- c("SKAT", "Burden")
test.stat.vals <- c("Score", "Wald", "Firth")
GetFamilyDistribution <- function(response.type) {
if (response.type == "Continuous"){
family = "gaussian"
} else if (response.type == "Dichotomous"){
family = "binomial"
} else {
msg = paste("Don't know how to deal with response type", response.type)
stop(msg)
}
return(family)
}
GetKinshipMatrix <- function(kinship.matrix){
cat('Loading Kinship Matrix:',kinship.matrix,'\n')
if(grepl('Rda',kinship.matrix,ignore.case=TRUE)){
kmatr = get(load(kinship.matrix))
}
else{
warning('Reading matrix file from text - will be dense format\n')
kmatr = as.matrix(read.csv(kinship.matrix,as.is=T,check.names=F,row.names=1))
}
cat('Loaded Kinship NROW:',NROW(kmatr),' NCOL:',NCOL(kmatr),'\n')
kmatr
}
cat('kinship.matrix',kinship.matrix,'\n')
cat('test.stat',test.stat,'\n')
cat('outcome.type',outcome.type,'\n')
cat('het_vars',het_varsIn,'\n')
cat('conditional',conditional,'\n')
if (!(test.stat %in% test.stat.vals)){
msg = paste("The requested test statistic:", test.stat, "is not available (Use Firth, Score, Wald!")
stop(msg)
}
.libPaths()
suppressMessages(library(SeqArray))
suppressMessages(library(SeqVarTools))
suppressMessages(library(GWASTools))
suppressMessages(library(gap))
suppressMessages(library(Matrix))
suppressMessages(library(plyr))
suppressMessages(library(gdsfmt))
suppressMessages(library(bdsmatrix))
suppressMessages(library(GENESIS))
suppressMessages(library(data.table))
## Setup
source("/genesis_wdl/pipelineFunctions.R")
sessionInfo()
if (covariate.string == "NA") {
covariates <- NULL
} else {
covariates <- split.by.comma(covariate.string)
}
cat('covariates:',paste(covariates, collapse = ", "),'\n')
cat('all terms:', paste(unique(c(outcome.name, covariates, het_varsIn)), collapse = ", "), '\n')
## phenotype
phenotype.data <- read.csv(pheno.file, header=TRUE, as.is=TRUE)
# add sex if provided, needed for X chrom MAF calcs
if( 'sex' %in% names(phenotype.data)){
gcol='sex'
# if only contains 'F', sex gets read as T/F
if (class(phenotype.data$sex) == 'logical') phenotype.data$sex = ifelse(phenotype.data$sex == FALSE,'F',phenotype.data$sex)
if(! all(phenotype.data$sex %in% c('F','M'))){
msg = paste("sex column must be coded M/F")
warning(msg)
#stop(msg)
}
}else{
if(! 'sex' %in% names(phenotype.data)) warning("A column labeled 'sex' coded M/F must be included when performing analyses on the sex chromosomes")
gcol = NULL
}
if(NCOL(phenotype.data) < 2) warning("Is the phenotype file a CSV? Too few columns from read.csv()")
cat('Input pheno N=',nrow(phenotype.data),'\n')
if(het_varsIn != 'NA'){
cat('prep pheno with het vars')
het_vars = het_varsIn
pheno <- reducePheno(phenotype.data, outcome.name, covariates=covariates,hetvars=het_vars, id=pheno.id, gender=gcol)
}else{
cat('prep pheno without het vars\n')
het_vars = NA
pheno <- reducePheno(phenotype.data, outcome.name, covariates=covariates, id=pheno.id, gender=gcol)
}
cat('Output pheno N=',nrow(pheno),'\n')
## Report dropped individuals
dropped.ids.selector <- !(phenotype.data[[pheno.id]] %in% row.names(pheno))
dropped.ids <- phenotype.data[[pheno.id]][dropped.ids.selector]
if (NROW(dropped.ids) != 0 ) {
cat("Dropped because of incomplete cases:", length(dropped.ids) )
}
# For GDS files
##f <- seqOpen('genotypefile')
#AAsample.ids <- seqGetData(f, "sample.id")
all.terms <- unique(c(outcome.name, covariates, het_vars,gcol))
#AApheno <- pheno[row.names(pheno) %in% sample.ids,na.omit(all.terms),drop=F]
cat('Output pheno after mergeing with Genos N=',nrow(pheno),'\n')
if(nrow(pheno) == 0){
msg = paste("Phenotype ID column doesn't match IDs in GDS")
stop(msg)
}
#subset to phenotyped samples
#AAseqSetFilter(f,sample.id = row.names(pheno))
# order pheno to the GDS subject order
#AAssample.ids <- seqGetData(f, "sample.id")
#AAspheno <- pheno[match(sample.ids,row.names(pheno)),,drop=F]
## Conditional analaysis
if(conditional != 'NA'){
f <- seqOpen(genotype.file)
sample.ids <- seqGetData(f, "sample.id")
pheno <- pheno[match(sample.ids,row.names(pheno)),,drop=F]
cond_snps = strsplit(conditional,',')[[1]]
cond_snps = gsub(' ','',cond_snps)
cond_snps = gsub('"','',cond_snps)
chr_array = seqGetData(f, "chromosome")
pos_array = seqGetData(f, "position")
allele_array = seqGetData(f, "allele")
allele_array = gsub(',',':',allele_array)
snp_array = paste(chr_array,pos_array,allele_array,sep=':')
cat('Conditioning on ',conditional,'...\n')
cidx = which(snp_array %in% cond_snps)
NCOND = length(cidx)
cat('Found ',snp_array[cidx], ' in GDS file\n')
if(NCOND < length(cond_snps)){
warning('NOT ALL CONDITIONAL SNPS FOUND IN GDS')
}
if(any(cidx)){
seqSetFilter(f,variant.sel=cidx, sample.id = row.names(pheno),verbose=FALSE)
ncol = NCOL(pheno)
pheno = cbind(pheno,as.data.frame(altDosage(f, use.names=FALSE)))
condheaders = paste0('csnp',1:NCOND)
colnames(pheno)[(ncol+1):(ncol+NCOND)] = condheaders
}else{
stop('Can not find snp ',conditional,' with position ',cpos,' to condition on in data file')
}
dropConditionalCases = NROW(pheno)-NROW(pheno[complete.cases(pheno),])
if(dropConditionalCases > 0){
cat('Warning: Dropping ',dropConditionalCases,' samples due to missing conditional genotype calls\n')
}
pheno = pheno[complete.cases(pheno),]
covariates[(length(covariates) + 1):(length(covariates) +NCOND)] <- condheaders
seqClose(f)
}
## Load KINSHIP matrix
## Kinship doesn't contain all samples
if(kinship.matrix != 'NO_KINSHIP_FILE'){
kmatr = GetKinshipMatrix(kinship.matrix)
pheno = pheno[row.names(pheno) %in% row.names(kmatr),,drop=F]
kmatr = kmatr[row.names(kmatr) %in% row.names(pheno),colnames(kmatr) %in% row.names(pheno)]
cat('Output pheno in Kinship N=',nrow(pheno),'\n')
kmatr = kmatr[match(row.names(pheno),row.names(kmatr)),match(row.names(pheno),colnames(kmatr))]
if(nrow(pheno) == 0){
msg = paste("Phenotype ID column doesn't match IDs in Kinship Matrix")
stop(msg)
}
if (!(identical(row.names(kmatr),row.names(pheno)))){
stop("Something is off problem with re-ordering")
}
## Get sample ids to check order
#AAseqSetFilter(f,sample.id = row.names(pheno))
#AAsample.ids <- seqGetData(f, "sample.id")
#AAif (!(identical(sample.ids,row.names(pheno)) && identical(row.names(kmatr),row.names(pheno)))){
#AA stop("Something is off problem with re-ordering")
#AA}
} else {
kmatr <- NULL
}
###################
## NULL MODEL
##################
print(class(kmatr))
## pheno does not need to be the same order as kmatr, but must have same dims at kmatr
cat('start fit....\n')
if (kinship.matrix == 'NO_KINSHIP_FILE'){
cat('Fitting unrelated model')
nullmod <- fitNullModel(pheno,
covars = covariates,
outcome = outcome.name,
family = GetFamilyDistribution(outcome.type))
}else{
# kmatr = as.matrix(kmatr)
if(het_varsIn == 'NA'){
cat('Fitting model with GRM ... \n')
nullmod <- fitNullModel(pheno,
covars = covariates,
outcome = outcome.name,
family = GetFamilyDistribution(outcome.type),
cov.mat = kmatr)
}else{
cat('Fitting model with GRM and het vars ...\n')
nullmod <- fitNullModel(pheno,
covars = covariates,
outcome = outcome.name,
group.var = het_vars,
family = GetFamilyDistribution(outcome.type),
cov.mat = kmatr)
}
}
save(nullmod,pheno,file= paste0(results.file, '.RData'))