Poly(A) selection vs rRNA depletion: guide for picking the right upstream method.

RNA-seq begins with a choice you cannot undo later: enrich polyadenylated transcripts or deplete rRNA and sequence what remains. That decision determines which molecules enter your libraries, how tolerant the workflow is to degraded or FFPE RNA, and which analyses will be most reliable. This guide explains how each strategy works, what it captures and misses, and how to choose based on organism, RNA quality, and the question you want to answer. ^{1 2}

What you will see in the data

Poly(A) selection removes most rRNA and many non-informative classes up front, so more reads map to annotated exons and statistical power for gene-level mRNA comparisons improves at the same depth. When RNA is fragmented, coverage tilts toward the 3′ end and long transcripts are undercounted because capture depends on an intact tail. rRNA depletion retains poly(A)+ and non-polyadenylated RNAs, so intronic and intergenic fractions increase. That “extra” signal is often informative: intronic reads track transcriptional changes while exonic reads integrate post-transcriptional processing, which lets you separate mechanisms when you model them together. ^{1 3}

Mechanisms in brief

Poly(A) selection. Oligo-dT hybridization captures RNAs with poly(A) tails, enriching mature eukaryotic mRNA and many polyadenylated lncRNAs while excluding most rRNA, tRNA, sn/snoRNA, and tail-less transcripts such as replication-dependent histone mRNAs.

rRNA depletion. Starting from total RNA, sequence-specific DNA probes hybridize to cytosolic and mitochondrial rRNAs and the hybrids are removed by RNase H or affinity capture; enzymatic post-cDNA options exist as well. Because selection targets rRNA sequences rather than a tail, the remaining pool includes both poly(A)+ and non-polyadenylated species, for example pre-mRNA, many lncRNAs, histone mRNAs, and some viral RNAs. Comparative work shows depletion is more resilient on fragmented and FFPE RNA.^{2 4 6}

When to use what

Choose poly(A) selection when the samples are eukaryotic and intact (for example RQS or RIN ≥ 7 or DV200 ≥ 50 percent) and your primary endpoint is gene-level changes in coding mRNA. Expect high exonic fractions and relatively low residual rRNA; if integrity is borderline, expect a visible 3′ bias and depressed counts for long transcripts. ²

Choose rRNA depletion when inputs are degraded or FFPE, when your question requires non-polyadenylated RNAs, or when cohort quality varies and you need one strategy that copes with mixed integrity. On compromised RNA, depletion usually preserves 5′ coverage better than poly(A) capture, and it keeps intronic signal that can be modeled jointly with exonic counts.^{2 4 6} For bacteria, poly(A) capture does not recover mRNAs because prokaryotic polyadenylation is sparse and often marks decay rather than stability, so depletion or targeted capture is standard. ⁵

Method selection table

Common pitfalls and fast fixes

Running poly(A) selection on heavily fragmented or FFPE RNA produces strong 3′ bias and under-representation of long transcripts. If the input sample is degraded, switch to depletion rather than raising depth. In depletion workflows, the dominant failure mode is probe mismatch in non-model organisms, which leaves high residual rRNA and wasted reads. Pilot a few samples and check percent rRNA and mapping profiles before scaling. ^{4 6}

Summary

The enrichment method defines the transcriptome you measure. Choose based on three filters: organism, RNA integrity, and whether the RNAs are polyadenylated. Intact eukaryotic RNA with a coding-mRNA question points to poly(A) selection. Degraded or FFPE inputs, any need for tail-less RNAs, or prokaryotic work point to rRNA depletion. Pick one method for a study and keep it constant.