collect_nf

collect.nf

#!/usr/bin/env nextflow

params.testdir = "/cephfs/covid/software/sam/nf-test/"

params.fofn_single = [params.testdir, 'samples.csv'].join('/')
single_manifest_ch = Channel
    .fromPath(params.fofn_single)  // open say, a CSV
    .splitCsv(header:true)         // split text stream into CSV records
    .map { row-> file(row.path) }  // coerce the record and return a file for each line
    .collect()                     // emit all items as one

process echo_fofn_single {

    input:
    file '*.txt' from single_manifest_ch

    output:
    publishDir path: "${params.testdir}", pattern: "out", mode: "copy", overwrite: true
    file "out"

    script:
    """
    cat *.txt > out
    """
}


params.fofn_group = [params.testdir, 'grouped_samples.tsv'].join('/')
group_manifest_ch = Channel
    .fromPath(params.fofn_group)
    .splitCsv(header:true, sep:'\t')
    .map { row-> tuple(row.group, file(row.path)) }  // coerce the record into a (group, file) tuple
    .groupTuple()                                    // return tuples of (group, [file1, ...fileN])

process echo_fofn_group {

    input:
    tuple(group, file(files)) from group_manifest_ch

    output:
    publishDir path: "${params.testdir}", pattern: "g*out", mode: "copy", overwrite: true
    file "g${group}.out"

    script:
    """
    cat ${files} > g${group}.out
    """
}

file1.txt

hello

file2.txt

world

file3.txt

hoot

file4.txt

meow

file5.txt

bang

g1.out

hello
world

g2.out

hoot
meow

g3.out

bang

grouped_samples.tsv

out

hello
world

samples.csv

path
/cephfs/covid/software/sam/nf-test/file1.txt
/cephfs/covid/software/sam/nf-test/file2.txt