How would you detect whether RNA degradation is driving differences between RNA-seq samples?

Why

RNA degradation does not shrink all genes equally. It distorts read coverage in a gene-specific way, so it can look like differential expression. The authors of the TIN method cite a study in which up to 56 percent of genes were differentially expressed because of degradation in vitro.

The RIN measured before library preparation is the usual check, but it scores the RNA as a whole, mostly from ribosomal peaks. TIN is computed from the RNA-seq reads themselves and scores how evenly each transcript is covered, then summarizes per sample. In poly(A) libraries, degraded RNA shows a clear sign, which the Conesa review describes: reads pile up at the 3' end of transcripts, because only fragments still attached to the poly(A) tail are captured.

In practice, plot gene-body coverage per sample, compute TIN, and check whether low values fall mostly in one condition. In the TIN paper, comparing 10 lower-integrity with 10 higher-integrity prostate cancer samples, calibrating counts with TIN reduced false positives and recovered meaningful pathways. When degradation is balanced across groups it mainly adds noise.

When it is confounded with treatment, no adjustment fully rescues the comparison.