nextflow script of RNA data analysis

Question

[Srinka]

This is the code I have written till mapping. It works fine till the trimming part as it takes 1 input. But throws an error while mapping. Please help in finding where I went wrong.

//rna_script.nf
nextflow.enable.dsl=2

/*
 * pipeline input parameters
 */
params.reads = "/home/user2/rnatoy/data/ggal/*.fq"
params.genome = "/home/user2/rnatoy/data/ggal/ggal_1_genome.fa"
params.outdir = 'results'

log.info """\
         R N A  P I P E L I N E    
         =============================
         genome: ${params.genome}
         reads : ${params.reads}
         outdir: ${params.outdir}
         """
         .stripIndent()


/*
 * Step 1. QUALITY CHECK
 */
process QUALITY {

    input :
    path( params.reads)



        script:
        """
        fastqc  ${params.reads} -o /home/user2/Dock/results

        """
}
/*
 * Step 2.TRIM
*/

process TRIM {

    input :
    path params.reads

    output :
    path '*trimmed.fq', emit : trimmed

        script:
        """
        trim_galore  ${params.reads} -o /home/user2/Dock/results/TRIMMED/trimmed.fq
        """

}

/*
 * Step 3. Builds the genome index required by the mapping process
 */
process buildIndex {
    input:
    path params.genome

    output:
    path 'genome.index*', emit : index

    """
    bowtie2-build --threads $task.cpus ${params.genome}  genome.index
    """
}

/*
 * Step 3. ALIGNMENT
 */
process mapping {     
    input:
    path TRIM.out
    path buildIndex.out 

    output:
    path "accepted_hits.sam"

    """
    bowtie2 -x buildIndex.out ${trimmed} -S accepted_hits.sam
    """
}

workflow {

    buildIndex(Channel.fromPath( params.genome ) )
    QUALITY(Channel.fromPath( params.reads ) )  
    TRIM(Channel.fromPath( params.reads ) )
    mapping(Channel.fromPath(TRIM.out, buildIndex.out ))


}

[Pierre Lindenbaum]

    trim_galore  ${params.reads} -o /home/user2/Dock/results/TRIMMED/trimmed.fq

this is really wrong.

trim_galore should use the input from its own process and the output shouldn't be a full path.

[Srinka]

Answer 1

[Pierre Lindenbaum]

there is not output named "out" for TRIM. It's named "trimmed"

path '*trimmed.fq', emit : trimmed

you want something like: TRIM.trimmed

Answer 2

[Ming Tommy Tang]

Unless you want to have some custom code, I suggest you start with https://nf-co.re/rnaseq

[ATpoint]

I encourage people writing their own workflows and understanding Nextflow logic. nf-core pipelines can be valuable, but by just running them you won’t learn a single piece of Nextflow.

[Ming Tommy Tang]

yes, you are right. If you want to learn it you should write it yourself. If it is a one-off thing that you have to do it quickly (as in industry), go to the dockerized workflow is the right choice.

Answer 3

[colindaven]

Paths in nextflow I see you are trying to output nextflow data to a hard coded path when trimming - don't to this, it will cause major problems, so let Nextflow handle input and output directories and paths for you as much as possible.

    trim_galore  ${params.reads} -o /home/user2/Dock/results/TRIMMED/trimmed.fq

So just do something like this

    trim_galore  ${params.reads} -o trimmed.fq

I would also learn how to put your reads in an input channel instead of passing ${params.reads} around everywhere. See great examples here:

https://nextflow-io.github.io/patterns/

[Srinka]

Thank you so much for this help but another error is thrown as shown in the picture -

The workflow was given as -

workflow {

mapping(Channel.fromPath('TRIM.out', 'buildIndex.out'))

}

The process is written as -

process mapping {     
    input:
    path 'TRIM.out'
    path 'buildIndex.out'


    """
    bowtie2 -x $buildIndex.out $TRIM.out -S accepted_hits.sam
    """

}

[colindaven]

Have a look at the complete Nextflow examples for RNA-seq - these deal with multiple inputs and outputs, which is the problem here.

https://github.com/CRG-CNAG/CalliNGS-NF/blob/master/modules.nf

workflow {

mapping(Channel.fromPath('TRIM.out', 'buildIndex.out'))

}

// I write my processes like this:

process mapping {     
    input:
    file trimmed_R1  // just try with Read1 until you get the hang of how it works, Read2 will complicate the pipeline
    file index    // this will only input 1 file called index, which is not sufficient as indices are mostly 6+ files

"""
bowtie2 -x $index $trimmed_R1 -S accepted_hits.sam
"""

}

[Srinka]

Thank you. Also, can you help me prioritize the command to run first? I am not able to find any command that suggests that.

[colindaven]

Sorry, not sure what you mean. I think it's worth looking at a lot of entry level Nextflow stuff to pass 1 file input and 1 file output through a workflow using Nextflow DSL2. Nothing fancy, just to gain experience. Or take an RNA-seq pipeline like the output I suggested, break it, and repair it - googling the error messages to get more experience. The Nextflow docs are pretty good and they provide lots of examples.

[colindaven]

Here is a youtube video of an RNA-seq pipeline being built in nf -