diff --git a/src/bin/kmerclust_load.R b/src/bin/kmerclust_load.R index d8afb0bbd294d4cb582913e3623ed2bf9430ade7..3185e83634e06023e8eaee9c1dec6630e741ce60 100644 --- a/src/bin/kmerclust_load.R +++ b/src/bin/kmerclust_load.R @@ -2,9 +2,15 @@ library(kmerclust) library(vroom) args <- commandArgs(trailingOnly = TRUE) -data <- vroom::vroom(args[2], show_col_types = FALSE) -format(object.size(data), units = "Mb") -annotation <- parse_annotation(args[1]) -count <- annotate_counts(annotation, data, args[3]) %>% dplyr::ungroup() %>% dplyr::select(-c("files")) +count <- vroom::vroom( + args[2], + show_col_types = FALSE, + col_names = c("kmer", "count_m", "count_f"), + skip = 1) %>% + mutate( + count_m = as.integer(count_m), + count_f = as.integer(count_f), + specie = as.factor(args[3]) + ) format(object.size(count), units = "Mb") save(count, file = paste0(args[3], ".Rdata"))