Hi #C0493FQQZ3P #CE8SSJV3N, I am interested in performing isoform analysis on short read data (150bp) using the edgeR package, following the example from the edgeR User’s Guide, section: Refer: 4.6 Differential transcript expression of human lung adenocarcinoma cell lines I did wanted to try using the

quant.sf

files from the

/salmon

run by referring edgeR user guide. I tried the below steps, but I see there are NA values in the Overdispersion. Probably

nf-core rnasplice

did not run with

bootstrap/gibbs replicates

, Salmon with inferential replicates: ? I there a way in the code to turn that on.

> setwd("/projects/RNA/salmon/")

> quant <- dirname(list.files(".", "quant.sf", recursive = TRUE, full.names = TRUE))

> print(quant)

[1] "./DD21_progenitors" "./DD39_progenitors" "./DD48_progenitors" "./DD98_progenitors"

> catch <- catchSalmon(paths = quant)

Reading ./DD21_progenitors, 427366 transcripts, 0 none samples
Reading ./DD39_progenitors, 427366 transcripts, 0 none samples
Reading ./DD48_progenitors, 427366 transcripts, 0 none samples
Reading ./DD98_progenitors, 427366 transcripts, 0 none samples

Try putting this in

salmon_quant.config

and use that with

-c

when you run the pipeline:

process {
    withName: 'SALMON_QUANT' {
        ext.args = '--numGibbsSamples 50'
    }
}

This process is used with different names, depending on context (

SALMON_QUANT_STAR

and

SALMON_QUANT_SALMON

), I hope it works with the original name. If not, we'll have to use a regex. Normally you would need to copy the

ext.args

from the original config (in this case in

conf/modules.config

), but there is no

ext.args

for this process. You can see if it worked by checking the command for the process as discussed before. 🤞

@Thomas Danhorn Thanks. Is my understanding correct? 1. Create a file named

salmon_quant.config

with the following content. This configuration targets both

SALMON_QUANT_STAR

and

SALMON_QUANT_SALMON

processes, adding the

--numGibbsSamples 50

argument.

process {
    withName: 'SALMON_QUANT_STAR' {
        ext.args = '--numGibbsSamples 50'
    }
    withName: 'SALMON_QUANT_SALMON' {
        ext.args = '--numGibbsSamples 50'
    }
}

2. Update my pipeline script

run_rnasplice.sh

script to include the custom configuration file using the

-c

option:

nextflow run /projects/toufim/nfcore_rnasplice/1_0_4/rnasplice \
--input /flashscratch/toufim/23_Palucka_003_004_samplesheet.csv \
--contrasts /flashscratch/toufim/23_Palucka_003_004_Contrastsheet.csv \
--outdir /flashscratch/toufim/ \
--star_index /projects/STAR_INDEX/STAR_Index_GENCODE/hg38_sjdbOver_150_merged_gtf/ref/ \
--fasta /projects/STAR_INDEX/STAR_Index_GENCODE/hg38_sjdbOver_150_merged_gtf/hg38.fa \
--gtf /projects/STAR_INDEX/STAR_Index_GENCODE/hg38_sjdbOver_150_merged_gtf/LUAD_v3.gencode_v44.merged.gtf \
--aligner star \
--gencode true \
--gtf_extra_attributes gene_id \
--rmats_paired_stats false \
--rmats_read_len 151 \
--sashimi_plot false \
--dexseq_dtu false \
-profile jax \
-c /flashscratch/singularity_cache_nfcore/jax.config \
-c /flashscratch/singularity_cache_nfcore/salmon_quant.config \
-resume

3. Run the script Ps: This how my current part of

jax.config

file looks, a question, instead of creating a new

salmon_quant.config

, does it allow to add point 1

withName

in this

jax.config

file so I can use one file only?

singularity {
    enabled    = true
    autoMounts = true
    cacheDir   = params.singularity_cache_dir
}
params {
    max_memory = 720.GB
    max_cpus   = 32
    max_time   = 336.h
}

process {
    withLabel: custom_mem {
        cpus = 32
        memory = 720.GB
        queue = 'compute'
    }
}

process {
    // Add the specific memory setting for the RMATS process
    withName: NFCORE_RNASPLICE:RNASPLICE:RMATS:RMATS_POST {
        memory = 100.GB
    }

    // Add the specific memory setting for the RMATS process
    withName: NFCORE_RNASPLICE:RNASPLICE:SALMON_QUANT_STAR {
        memory = 100.GB
    }

    // Add the specific memory setting for the RMATS process
    withName: NFCORE_RNASPLICE:RNASPLICE:ALIGN_STAR:BAM_SORT_STATS_SAMTOOLS:SAMTOOLS_INDEX {
        memory = 100.GB
    }
}

I think your proposed config should work. I think the way I wrote it should work as well (as the actual process is called

SALMON_QUANT

, the other names are aliases to allow for different configs), or you could use a regex to capture both of the names in a single

withName:

clause. It's a question of taste, use whatever works for you.

... does it allow to add point 1

withName

in this

jax.config

file ...

You certainly can do that, the question is, should you? I personally would not combine these files. The

jax.config

is a general config file that tells an nf-core pipeline about your cluster. You could potentially use that with any Nextflow/nf-core pipeline. In fact, I would remove the rnasplice-specific clauses from

jax.config

and put them into a separate file. E.g., you could rename the

salmon_quant.config

to

rnasplice.config

and put everything specific to that pipeline in there, unless there might be cases where you want to run the pipeline with one part, but not the other part of that config, in which case separate files are better.

@Thomas Danhorn Thank you very much for the guidance. Yes, you are right, it’s better to create separate files to add the rnasplice processes. I created a separate

rnasplice.config

file and added the process with regex (see below), and it worked. The other workaround (having one config file) also worked. Initially, I thought having two -c config files would override one another, so I considered including everything together in one config file.

process {
    // regex Match any process name that starts with 'SALMON_QUANT'
    withName: /SALMON_QUANT_.*/ {
        ext.args = '--numGibbsSamples 50'
    }
}

If you are changing the same thing (e.g. memory for a specific process) in two config files, they will most likely override each other for that, but otherwise the contents of both files are used.

This is noted.