Input for Deseq2 for downstream analysis

Question

[وفاء]

Hi everyone,

I ran the nf-core/rnaseq pipeline in the cluster for three studies separately (because they huge data), and I got some files as the output for each one of them:

salmon.merged.gene_counts_length.tsv (29744 genes)
salmon.merged.gene_counts_length_scaled.tsv (29744 genes)
salmon.merged.gene_counts_scaled.tsv (29744 genes)
salmon.merged.gene_counts.tsv (29744 genes)
salmon.merged.gene_temp.tsv (29744 genes)

Then I combine all the outputs ("salmon.merged.gene_counts_length_scaled.tsv") of the three studies into one matrix as input for Deseq2. So, however, I am not sure which files I need to use and combine since I read that the input for Deseq2 should be unnormalized and the only unnormalized files are the "salmon.merged.gene_counts.tsv".

So, which files should I use, and why? any help will be appreciated.

Answer 1

[ATpoint]

The file salmon.merged.gene_counts_length_scaled.tsv represents raw counts but corrected for average transcript length per gene. It is the output of running tximport with the lengthScaledTPM option. I find this a good choice because you can feed this file into any downstream analysis, be it DESeq2, limma or others.

See https://bioconductor.org/packages/release/bioc/html/tximport.html for tximport details.