How to Remove rRNA from RNA-Seq Libraries: A Method Comparison
Team Dynamic Matrices | 18 April 2026
Ribosomal RNA is most of the RNA in any cell. Getting it out of a sequencing library is a standard step, and the method you choose decides what fraction of your reads are usable, which transcripts you can see, and whether the workflow survives degraded or low-input samples. Here are the four approaches, how each works, and where each belongs. One of the four is ours.
Method 1: Probe hybridisation with bead pulldown
Biotinylated DNA or RNA probes tiled across the rRNA sequences hybridise to rRNA; streptavidin beads pull the hybrids out; the supernatant is your depleted RNA. This was the Ribo-Zero mechanism and remains common.
Strengths: Well validated on intact RNA from model organisms; large literature.
Weaknesses: Probes are designed against a reference. Sequence variants in non-model organisms escape. Fragments that do not span a probe site escape, so efficiency drops on FFPE and degraded input (Adiconis et al., Nature Methods, 2013). Bead capacity caps how much rRNA can be removed per reaction. Herbert et al. (BMC Genomics, 2018) found substantial site-to-site variability across kits.
Method 2: RNase H digestion
DNA oligonucleotides tiled across rRNA hybridise to it; RNase H degrades the RNA strand of the resulting DNA:RNA hybrids; DNase then removes the oligos. NEBNext rRNA Depletion and KAPA RiboErase use this.
Strengths: No beads, scalable, works well on intact RNA, cheaper than probe pulldown.
Weaknesses: Depends on hybrid formation, so efficiency drops with fragmented or low-input RNA where short fragments do not stably hybridise. Two enzymatic steps with their own lot and cold-chain dependence. Oligo panels are still organism-specific.
Method 3: Poly(A) selection
Oligo-dT on beads captures polyadenylated mRNA; everything else, including rRNA, is discarded. Not a depletion of rRNA but an enrichment of mRNA.
Strengths: Simple, cheap, very low rRNA in the library, good for intact eukaryotic mRNA quantification.
Weaknesses: Discards non-polyadenylated RNA: most lncRNA, histone mRNAs, circRNA, bacterial and many viral transcripts. Fails on fragmented RNA because the poly(A) tail is separated from the transcript body, giving strong 3' bias (Zhao et al., Scientific Reports, 2018). Unsuitable for FFPE.
Method 4: Sequence-selective phase separation (riboLASSO)
A library of short DNA catcher strands complementary to rRNA is attached to a soluble linear polymer. When catchers hybridise to rRNA, and to each other through the rRNA, the polymer network crosslinks and undergoes a phase transition, condensing into a pellet that is spun down. The supernatant is depleted RNA. No beads, no enzymes.
Strengths: Capture is by sequence, not by a fixed probe site, so short fragments carrying any catcher-complementary region are captured; efficiency holds on degraded and low-input RNA. Capacity scales with polymer amount, about 8 to 20 times a bead. Reagents are enzyme-free and stable for about a year at 4 °C. Over 98 percent depletion on human total RNA in our benchmarks; reagent cost about one euro per sample. Catcher libraries can be designed for any organism whose rRNA sequence is known, and we have a bacterial library in development (Dynamic Matrices, Angewandte Chemie, 2025; patent WO 2025/196123A1).
Weaknesses: Newer method with a shorter validation history than the others. Currently released for human; other organisms need a catcher library.
Side by side
| Probe pulldown | RNase H | Poly(A) selection | riboLASSO | |
|---|---|---|---|---|
| Mechanism | Hybridise + bead | Hybridise + digest | Oligo-dT bead | Hybridise + phase separation |
| Enzymes | No (beads) | Yes, two | No | No |
| Beads | Yes | No | Yes | No |
| Intact RNA efficiency | High | High | High (mRNA only) | High |
| Degraded / FFPE | Drops | Drops | Fails | Holds |
| Low input | Surface losses | Hybridisation-limited | Poor | Holds to low ng |
| Non-poly(A) RNA retained | Yes | Yes | No | Yes |
| Non-model organisms | Needs panel | Needs panel | Eukaryotes only | Needs catcher library |
| Reagent stability | Cold | Cold | Cold | About a year at 4 °C, no freezer |
| Reagent cost per sample | High | Medium | Low | Low |
| Validation history | Long | Long | Long | Short |
Which to use
- Intact RNA, model organism, mRNA quantification only: poly(A) selection is cheapest and fine.
- Intact RNA, model organism, whole transcriptome: RNase H or probe pulldown, both well validated; riboLASSO if cost, enzyme logistics or capacity matter.
- Degraded, FFPE or low-input samples: a method whose efficiency does not depend on intact probe sites or enzymatic completion. This is where riboLASSO is strongest and where we suggest a side-by-side.
- Non-model organisms: whichever method has a panel or catcher library for your organism, or can make one. Methods not tied to a pre-validated commercial panel transfer better.
References
- Adiconis X et al. Comparative analysis of RNA sequencing methods for degraded or low-input samples. Nat Methods 10, 623–629 (2013).
- Herbert ZT et al. Cross-site comparison of ribosomal depletion kits for Illumina RNAseq library construction. BMC Genomics 19, 199 (2018).
- Zhao S et al. Evaluation of two main RNA-seq approaches for gene quantification in clinical RNA sequencing: polyA+ selection versus rRNA depletion. Sci Rep 8, 4781 (2018).
- Morlan JD, Qu K, Sinicropi DV. Selective depletion of rRNA enables whole transcriptome profiling of archival fixed tissue. PLOS ONE 7, e42882 (2012).
- Dynamic Matrices. LASSO: sequence-selective biomolecule isolation by programmable polymer phase separation. Angew Chem Int Ed (2025); patent WO 2025/196123A1.
FAQs
Some workflows do poly(A) selection then a depletion step to remove residual rRNA, or two depletion rounds. Each step loses material; only combine when one method alone is demonstrably insufficient.
Not poly(A) selection. Probe or RNase H panels exist for common bacteria; for mixed communities, a method that can be designed against many rRNA sequences at once is needed. We are building a bacterial riboLASSO library for this reason.
qPCR for 18S or 28S versus a housekeeping mRNA before and after depletion. It takes an hour and it is the cheapest QC step in the workflow.
Ribosomal RNA makes up roughly 80 to 90 percent of total RNA, so without removal most sequencing reads would map to rRNA you then discard. rRNA removal, also called rRNA depletion, takes it out before library preparation; poly(A) selection is the alternative route, enriching mRNA instead. See our rRNA depletion guide and rRNA depletion vs poly(A) selection.
Not poly(A) selection: fragmentation separates the tail from the transcript body, so coverage collapses to a 3'-biased fraction. Probe pulldown and RNase H lose efficiency on fragments that do not span a probe site or hybridise stably (Adiconis et al., Nature Methods, 2013). A method that captures by sequence rather than a fixed probe site, such as riboLASSO, is designed for fragmented input, and we suggest a side-by-side on your own samples. Details: rRNA depletion for degraded RNA and FFPE RNA extraction challenges.
Choose among four approaches by RNA integrity, input and organism: probe hybridisation with bead pulldown, RNase H digestion, poly(A) selection (which enriches mRNA rather than removing rRNA) and sequence-selective phase separation. The table above compares them. Check the result by qPCR for 18S or 28S before and after depletion, and see riboLASSO vs other rRNA depletion kits.