An R script for FOCUS fine-mapping analysis

#!/usr/bin/env Rscript
slurm_arrayid <- Sys.getenv('SLURM_ARRAY_TASK_ID')
job.id <- as.numeric(slurm_arrayid)

#export PATH=$PATH:/gpfs/research/chongwu/shared/software

#job.id = 4

library(data.table)
library(RSQLite)
library(tidyverse)

args <- commandArgs(TRUE)
job <- (eval(parse(text = args[[1]])))
job <- as.numeric(job)


# prepare the indx
ref_ld = "/gpfs/research/chongwu/shared/1000Genomes/1000G.EUR.ALLSNP.QC.CHR"
loci = "/gpfs/research/chongwu/shared/LDetect_LD_regions/EUR/"
sumstat = "/gpfs/research/chongwu/Chong/Lang/AD_TWAS/processed_data.txt" # generated from munge.R

#significant gene list
twasresInput = "gene_AD.csv"
twasres = as.data.frame(fread(twasresInput))
#twasres[,"pvalue"] = as.numeric(twasres[,"pvalue"])


#job = 1

tissueall = c("Brain_Amygdala","Brain_Anterior_cingulate_cortex_BA24","Brain_Caudate_basal_ganglia","Brain_Cortex","Brain_Hippocampus","Brain_Hypothalamus","Brain_Nucleus_accumbens_basal_ganglia","Brain_Putamen_basal_ganglia","Brain_Substantia_nigra","Pituitary")

parentdir = "/gpfs/research/chongwu/Chong/Lang/AD_TWAS/"


if(job<=10) {
  
  tissue = tissueall[job]
  weights = paste("/gpfs/research/chongwu/shared/TWAS_JTI/UTMOST_",tissue,".db",sep="")

  sqlite.driver <- dbDriver("SQLite")
  db <- dbConnect(sqlite.driver,dbname = weights)

  dbListTables(db)
  weights =  dbReadTable(db, "weights")
} else {
    
    weights = NULL
    extra = NULL
    for (j in 1:length(tissueall)) {
        
        weight = paste("/gpfs/research/chongwu/shared/TWAS_JTI/UTMOST_",tissueall[j],".db",sep="")

        sqlite.driver <- dbDriver("SQLite")
        db <- dbConnect(sqlite.driver,dbname = weight)

        dbListTables(db)
        tmp =  dbReadTable(db, "weights")
        tmp$name = paste(tmp[,"gene"],"_",tissueall[j],sep="")
        weights = rbind(weights,tmp)
        
        tmp =  dbReadTable(db, "extra")
        tmp$name = paste(tmp[,"gene"],"_",tissueall[j],sep="")
        extra = rbind(extra,tmp)
    }
    

    dt <- data.table(extra)
    extra2 = dt[ , max(pred.perf.R2), by = gene]
    extra2 = as.data.frame(extra2)
    id = paste(extra2[,1],extra2[,2],sep="_")
    
    id2 = paste(extra[,"gene"],extra[,"pred.perf.R2"],sep="_")
    
    extra3 = extra[id2 %in% id,]
    
    extra3 = extra3[!duplicated(extra3[,"gene"]),]
    
    length(unique(weights$name ))
    weights = weights[weights$name %in% extra3$name,]
    
    tissue = "ALL"
}


outd = paste(parentdir,tissue,sep="")
weightssave = paste(parentdir,tissue,"_weights.txt",sep="")
saveprefix = paste(tissue,"_UT",sep="")
genelist = paste(parentdir,tissue,"_UT_gene_list.txt",sep="")


#tmp = dbReadTable(db, "construction")

#1. wgtlist
#rsid            gene      weight ref_allele eff_allele
#1 rs141364387 ENSG00000169583 -0.19646764          C          T


# gene linformation wgtlist0
#gene.inf = readRDS("/gpfs/research/chongwu/shared/GTEX8/gencode.v26.hg19.genes.rds")
#gene.inf$gene = gsub("\\..*","",gene.inf$gene_id)
#gene.inf = gene.inf[!duplicated(gene.inf$gene),] #10321
#gene.inf = gene.inf[,c("V4","V5","V1","gene")]

#gene.inf = as.data.frame(fread("ENSEMBL_GRch37_gene_list.txt"))
#gene.inf = gene.inf[,c(5,6,4,1)]
#colnames(gene.inf) = c("left","right","chr","gene")

#gene.inf = as.data.frame(fread("gencode.v19.exclude_pseudogene-1.list"))
#gene.inf[,"V5"] = gsub("\\..*","",gene.inf[,"V5"])
#gene.inf = gene.inf[,c(3,4,1,5)]
#colnames(gene.inf) = c("left","right","chr","gene")

# The above different versions do not cover all the genes

gene.inf = as.data.frame(fread("/gpfs/research/chongwu/shared/gencode.v19.annotation.gtf.gz"))
gene.inf = gene.inf[gene.inf[,3] %in% c("gene"),]

genetmp = gene.inf[,9]

tmp = gsub("\";.*","",genetmp)
tmp2 = gsub("gene_id \"","",tmp)
tmp2 = gsub("\\..*","",tmp2)
gene.inf$gene = tmp2

tmp = gsub(".*gene_name \"","",genetmp)
tmp = gsub("\";.*","",tmp)

gene.inf$name = tmp

gene.inf = gene.inf[,c(4,5,1,10)]
colnames(gene.inf) = c("left","right","chr","gene")

#gene.inf = gene.inf[gene.inf[,1]==paste0("chr",chr.id),]
dim(weights)
length(unique(weights$gene))
weights = weights[weights$gene %in% gene.inf$gene,]
dim(weights)
gene.inf = gene.inf[gene.inf$gene %in% weights$gene,]
dim(gene.inf)


write.table(weights,weightssave,col.names=TRUE,row.names=FALSE,quote=FALSE)

write.table(gene.inf,genelist,col.names=TRUE,row.names=FALSE,quote=FALSE)

#2. wgtlist0
#       left     right  chr            gene
#1 169818772 169863408 chr1 ENSG00000000457
#2 169631245 169823221 chr1 ENSG00000000460

gene.inf2 = gene.inf[gene.inf[,"gene"] %in% twasres[,"gene"],]

gene.inf2[,"chr"] = gsub("chr","",gene.inf2[,"chr"])

usedIndx = matrix(NA,dim(gene.inf2)[1],2)

for(i in 1:dim(gene.inf2)[1]) {
    tmp.loci = fread(paste(loci, "fourier_ls-chr",gene.inf2[i,"chr"],".bed",sep=""))
    tmp.loci = as.data.frame(tmp.loci)
    
    usedIndx[i,1] = gene.inf2[i,"chr"]
    
    tmpIndx = 1:dim(tmp.loci)[1]
    usedIndx[i,2] = tmpIndx[tmp.loci[,2]  < gene.inf2[i,1] & tmp.loci[,3] > gene.inf2[i,1]  ]
}


my_data <- as_tibble(usedIndx)
usedIndx = my_data %>% distinct()
usedIndx = as.data.frame(usedIndx)

chr.id = usedIndx[job.id,1]
loci.indx = usedIndx[job.id,2]


system(paste("Rscript FOGS.R --refld ",ref_ld," --outd ",outd," --loci ",loci," --weights ", weightssave," --genelist ",genelist, " --sumstat ",sumstat," --saveprefix ",saveprefix," --chr_id ",chr.id, " --locus_id ",loci.indx,sep=""))