Using SED command in Nextflow script

Question

I was wondering if someone can help me figure out this error message. I’m at the end of my Nextflow pipeline and I want to change the header in a FASTA file and it works when I use this command:

sed -i "s/^>/>${assembly_id}_${gene_id}_/g" "${assembly_id}/${gene_id}/${assembly_id}_${gene_id}.fasta"

but it prints out much more than I actually need (i.e., “goodID_extra_extra_extra”). That’s just part of the file name (it includes what assembler was used) but I only want it to add “goodID_” to the header.

I tried using this command:

sed -i "s/^>/>${assembly_id%%_*}_${gene_id}_/g" "${assembly_id}/${gene_id}/${assembly_id}_${gene_id}.fasta"

but I get this error message:

N E X T F L O W  ~  version 21.10.6
Launching `geneTargets.nf` [amazing_elion] - revision: b875027a52
Script compilation error
- file : /Users/rimo/Desktop/AWS/nextflow_work/PROJECTS/gene_targets/geneTargets.nf
- cause: Unexpected input: '{' @ line 94, column 32.
   process publish_gene_sequences {
                                  ^

1 error

Can someone explain why this is happening? If this is something I can’t do without breaking the script that’s okay the header will still do it’s purpose I just wanted it to be clearer/less noise.

EDIT: this is the process block

process publish_gene_sequences {
publishDir params.publish_dir
input:
    tuple val(assembly_id), val(gene_id), path("gene.fasta")
output:
    path("${gene_id}/${assembly_id}_${gene_id}.fasta")
script:
    """
    mkdir -p ${gene_id}
    cp gene.fasta "${gene_id}/${assembly_id}_${gene_id}.fasta"
    sed -i "s/^>/>${assembly_id}_${gene_id}_/g" "${gene_id}/${assembly_id}_${gene_id}.fasta"
    """
 }

Answer 1

Inside your script's command string, assembly_id is actually a java.lang.String so trying to manipulate it with a Bash operation (i.e. using ${assembly_id%%_*}) won't produce the expected result. Here's one way to return a substring up to the first underscore; note that we can use single quotes around our sed expression:

process publish_gene_sequences {

    publishDir params.publish_dir

    input:
    tuple val(assembly_id), val(gene_id), path("gene.fasta")

    output:
    path("${gene_id}/${assembly_id}_${gene_id}.fasta")

    script:
    def assembly = assembly_id.substring(0, assembly_id.indexOf('_'))

    """
    mkdir "${gene_id}"
    cp gene.fasta "${gene_id}/${assembly_id}_${gene_id}.fasta"
    sed -i 's/^>/>${assembly}_${gene_id}_/g' "${gene_id}/${assembly_id}_${gene_id}.fasta"
    """
}

Note also that in the above example, copying the file can be avoided by applying the expression directly and redirecting stdout:

script:
def assembly = assembly_id.substring(0, assembly_id.indexOf('_'))

"""
mkdir "${gene_id}"
sed \\
    -e 's/^>/>${assembly}_${gene_id}_/g' \\
    gene.fasta \\
    > "${gene_id}/${assembly_id}_${gene_id}.fasta"
"""