Generate nanopolish index in parallel

generate_nanopolish_index.R

#!/usr/bin/env Rscript
library(fs)
library(rhdf5)
library(parallel)
library(purrr)
library(stringr)
library(dplyr)
library(tidyr)

index_fast5 <- function(fastq_file, fast5_dir) {
    stopifnot(file_exists(fastq_file))
    
    f5_files <- path_real(
        dir_ls(fast5_dir, glob="*.fast5")
    )
    
    stopifnot(length(f5_files) > 0)
    
    names(f5_files) <- f5_files
    
    x <- mclapply(
        f5_files,
        partial(h5ls, recursive = FALSE),
        mc.cores = 128L # OS begins complaining when I go up to 512 processes
    )
    
    # this creates a table with almost the same information as readdb files
    # just need to trim the paths and read names
    index <- bind_rows(x, .id = "fast5_file") %>%
        mutate(read = str_remove(name, "^read_")) %>%
        select(read, fast5_file)
    
    readdb_name <- paste0(fastq_file, ".index.readdb")
    write_tsv(index, path = readdb_name, col_names = FALSE)
}

args <- commandArgs(trailingOnly = TRUE)

fastq_file <- args[1]
fast5_dir <- args[2]

index_fast5(fastq_file, fast5_dir)