Gene or Exon Counts for DESeq2 in rRNA-Depleted RNA-seq?

Question

[n@4j]

Hello everyone,

I have rRNA-depleted RNA-seq samples (without Oligo(dT) bead selection) that were counted at both exon and gene levels using featureCounts after alignment with STAR. However, the results differ:

a) Exon-level counting: assigned 40–60%, unassigned no features 40–60%, unassigned ambiguity <5%
b) Gene-level counting: assigned >80%, unassigned no features <10%, unassigned ambiguity <10%

Based on my study design, which count type should be used as input for DESeq2: exon-level counts or gene-level counts, and what would be the reasoning behind that choice?

Answer 1

[GenoMax]

People generally do the analysis at gene level. Unless they are interested in transcript/isoform level analysis.
Why Are Transcripts Differentially Expressed but Not Genes in RNA-seq Analysis? — Confusion Around Transcript-Level Differential Expression
and
What is the difference between deferentially expressed genes and deferentially expressed transcripts? .

[n@4j]

Thank you for your reply. I’m not really sure what you mean by “people generally do the analysis.” Are you referring specifically to rRNA-depleted RNA-seq analysis, or RNA-seq analysis in general? In another post,After read mapping, count exons or genes?, someone suggested using exon for RNA-seq, so now i am confused. Appreciate it if you could explain this further.

[GenoMax]

I was referring to RNAseq analysis in general. Most times people are interested in differentially expressed "genes" level analysis and not differential transcript analysis. You would need to keep exon level counts if you were doing differential transctipt analysis. Normally the exon counts are summarized to the gene level to generate a single expression value for each gene.