process_genotype.R

library(snpStats)
library(rrBLUP)
library(sommer)

args <- commandArgs(trailingOnly = TRUE)
train_path <- args[1]
test_path <- args[2]

geno_Train <- read.csv(train_path, header = TRUE, stringsAsFactors = FALSE)
geno_Test <- read.csv(test_path, header = TRUE, stringsAsFactors = FALSE)

sample_ids_Train <- geno_Train[, 1]
sample_ids_Test <- geno_Test[, 1]

geno_Train <- as.matrix(geno_Train[, -1])
geno_Test <- as.matrix(geno_Test[, -1])

additive_Train <- A.mat(geno_Train)
dominance_Train <- D.mat(geno_Train)
additive_Test <- A.mat(geno_Test)
dominance_Test <- D.mat(geno_Test)

write.csv(data.frame(Sample_ID = sample_ids_Train, additive_Train), "Train_additive.csv", row.names = FALSE)
write.csv(data.frame(Sample_ID = sample_ids_Train, dominance_Train), "Train_dominance.csv", row.names = FALSE)
write.csv(data.frame(Sample_ID = sample_ids_Test, additive_Test), "Test_additive.csv", row.names = FALSE)
write.csv(data.frame(Sample_ID = sample_ids_Test, dominance_Test), "Test_dominance.csv", row.names = FALSE)