In byandell/DOread: Read DO data

knitr::opts_chunk$set(echo = TRUE)

R Markdown

Convert from save/load to saveRDS/readRDS. Should improve access but will not change size.

library(readr)
library(qtl2geno)
dirpath <- file.path("~/Documents/Research/attie_alan/DO", "data")
datapath <- file.path(dirpath, "DerivedData")
wavepath <- file.path(datapath, "wave5")

Precomputed peaks

load(file.path(wavepath, "peaks.RData"))
saveRDS(peaks, file.path(datapath, "peaks.rds"))

Precomputed analyses

analysis_tbl <- read.csv(file.path(wavepath, "analyses.csv"),
                         stringsAsFactors = FALSE)
saveRDS(analysis_tbl, file.path(datapath, "analyses.rds"))

Physical map

load(file.path(wavepath, "probs_pmap.RData"))
saveRDS(pmap, file.path(datapath, "pmap.rds"))

Genotype probabilities (slow)

if(file.exists(file.path(wavepath, 
               "GM_Attie_allele_call_haploprobs_4qtl2_wave4.Rdata"))) {
  ## Genotype probabilities ("probs") [very large]
  load(file.path(wavepath, 
                 "GM_Attie_allele_call_haploprobs_4qtl2_wave4.Rdata"))
  ## wave3: "GM_Attie_allele_call_haploprobs_4qtl2.RData"
  ## physical map of the markers in the probs array
  saveRDS(probs, file.path(datapath, "probs.rds"))
  for(chr in names(probs$map)) {
    cat(chr, "\n")
    saveRDS(probs[,chr], 
            file.path(datapath, paste0("probs_", chr, ".rds")))
  }
} else {
  if(file.exists(file.path(wavepath, "probs_1.rds"))) {
    for(chr in names(pmap)) {
      cat(chr, "\n")
      probs <- readRDS(file.path(wavepath, paste0("probs_", chr, ".rds")))
      m <- match(names(probs$map[[chr]]), names(pmap[[chr]]))
      if(any(is.na(m)))
        stop("missing chr names")
      probs$map[[chr]] <- pmap[[chr]][m]
      saveRDS(probs[,chr], 
              file.path(datapath, paste0("probs_", chr, ".rds")))
    }
  }
}

Kinship matrices ("loco" method) ("K")

load(file.path(wavepath, "kinship_qtl2.RData"))
saveRDS(K, file.path(datapath, "kinship.rds"))

Covariates

load(file.path(wavepath, "covar.RData"))
saveRDS(covar, file.path(datapath, "covar.rds"))

Phenotypes

for(pheno_type in c("clin","gutMB")) {
  tmp <- paste("pheno", pheno_type, sep = "_")
  tmp2 <- paste(tmp, "dict", sep = "_")
  load(file.path(wavepath, paste0(tmp, ".RData")))
  saveRDS(get(tmp), file.path(datapath, paste0(tmp, ".rds")))
  saveRDS(get(tmp2), file.path(datapath, paste0(tmp2, ".rds")))
}