kmerclust_clust.R

library(kmerclust)
library(tidyverse)
args <- commandArgs(trailingOnly = TRUE)
print(args)

load(file = paste0(args[1], ".Rdata"))

res <- count %>%
  dplyr::select(count_m, count_f) %>%
  mutate(
    count_m = round(count_m),
    count_f = round(count_f)
  ) %>%
  dplyr::filter(count_m + count_f > 0) %>%
  as.matrix() %>%
  em_bipoiss_clust(nbatch = 100, max_iter = 1000)

save(res, file = paste0(args[1], "_clust_", args[3], ".Rdata"))

#count %>%
#  mutate(
#    count_m = round(count_m),
#    count_f = round(count_f)
#  ) %>%
#  dplyr::filter(count_m + count_f > 0) %>%
#  dplyr::mutate(chromosome = res$chromosome) %>%
#  dplyr::select(name, chromsome) %>%
#  write_csv2(file = paste0(args[1], "_clust_", args[3], ".csv"))