RNA-seq library types polyA and ribodepletion

A library kit chooses which RNA molecules are allowed to become reads. Poly(A) selection keeps RNA that binds an oligo-dT stretch. Ribosomal depletion, often called ribodepletion, removes the ribosomal RNA that would otherwise dominate the sequencer and leaves most of everything else. Those are different biological questions wearing similar file formats. This comparison is for anyone about to lock a library type on a sample sheet. The surrounding path is from cells to a gene expression result. A public orientation to short-read chemistry is the Illumina sequencing overview. It describes a technology class, not a kit recipe for your organism.

The decision, before the adapter is ligated

Write the molecules you need to see. Mature coding transcripts in intact eukaryotic cells are a poly(A) question. Bacterial expression, many non-coding RNAs, histone mRNAs, pre-mRNA, and RNA that has already been fragmented are usually a ribodepletion question. If you are unsure, look at a pilot integrity trace and at whether the organism even makes stable poly(A) tails on its mRNA. Changing your mind after sequencing means a new library. Software cannot put back the molecules the selection threw away.

The mRNA sequencing enquiry reference is where that choice can be discussed as an enquiry. It is a reference for the conversation. The comparison below is the science you should already be able to state when you write the quote request.

Poly(A) capture is a physical selection

Eukaryotic mRNA is transcribed, spliced and given a poly(A) tail that affects stability and translation. Oligo-dT beads or membranes bind that tail and wash away ribosomal RNA, transfer RNA and much of the non-polyadenylated fraction. The bound material is enriched for mature mRNA, which is why differential expression of coding genes in a clean cell-line experiment is often done this way. Sequencing depth then goes into those transcripts instead of into ribosomes.

The selection has edges. Histone mRNAs are a classic non-polyadenylated coding class and are largely lost. Many long non-coding RNAs are lost or only partly kept. Nascent pre-mRNA is kept only to the extent that it already carries a tail, so intronic signal is lower than in a ribodepleted library. Small RNAs are not this library at all. They need a size-selected chemistry, sketched in small RNA sequencing overview.

Degraded RNA fails oligo-dT capture for a physical reason. Fragmentation and exonuclease attack shorten or remove the tail, and the broken body of the transcript washes away. Formalin-fixed material and any sample that sat warm often look "fine" by mass and then yield almost nothing after poly(A) selection, or yield a biased subset of whatever tails survived. The upstream defence is protecting RNA during extraction. The library choice comes after that, not instead of it.

Bacteria are a harder mismatch. In bacteria, polyadenylation is generally a mark that encourages decay, not the stable tail of a mature eukaryotic message. An oligo-dT selection will not represent a bacterial transcriptome in the way a mammalian biologist expects. Ribodepletion with probes aimed at that species' ribosomal RNA is the usual class of library.

Ribodepletion is a subtraction

Ribosomal RNA can be most of the mass in a total RNA extract. If you sequence total RNA without removing it, the read budget is spent on a few ribosomal genes. Depletion methods hybridise probes to ribosomal sequences and digest or pull those hybrids away, or they use related enzymatic schemes. What remains includes mRNA, pre-mRNA and a share of non-coding RNA. Intronic and intergenic signal rises relative to a poly(A) library. That is useful when you care about immature transcripts, about genes with poor tails, or about organisms that are not polyadenylated like yeast and animals.

Degraded samples often work better here because the method does not ask for a tail. Fragmented coding sequence can still be captured as long as it is not the ribosomal target of the probes. "Better" still has a limit. Very short fragments, cross-linked extracts and heavy DNA contamination produce their own biases. Report an integrity measure beside the library choice so a later reader knows you were sequencing fragments on purpose.

Leftover ribosomal RNA is the control that tells you the subtraction worked. A library with a huge ribosomal fraction is a failed depletion even if the FASTQ files are large. Probe sets are organism-specific. A human ribosomal probe set will not correctly deplete a divergent microbe, and a mixed infection sample can leave the unmatched ribosomal RNA in place. Name the probe class and the species you believe are in the tube.

Strandedness sits beside the selection

Whether the library remembers the original strand is a separate choice. A poly(A) library can be stranded or not. A ribodepleted library can be stranded or not. Unstranded data collapses sense and antisense into one count. That is acceptable for some gene-level questions in well-annotated coding genes, and it is a poor fit when antisense transcription or overlapping genes are the point. State strandedness in the same sentence as the selection. Do not let a kit name imply it.

Read length, paired ends and the amount of input are further choices. Low-input amplification adds its own bias and is a reason to keep biological replicates honest, which is the subject of biological versus technical replicates. None of those choices repair the wrong selection.

Calling a missing fraction a biological zero

The sharpest failure is interpretive. A poly(A) dataset that shows no histone mRNA, no bacterial transcript and little intronic signal may be behaving exactly as designed. Writing "this gene is not expressed" because a non-polyadenylated transcript is absent is a library error described as biology. The matching failure for ribodepletion is to treat leftover ribosomal reads, or a surge of intronic signal from nuclei, as a mature cytoplasmic mRNA change. Look at the genomic distribution of reads before you lock the story.

DNA contamination consumes reads in both chemistries and can look like intergenic expression. A DNase step and a sense of strandedness, if you have it, help. Cross-sample comparisons require the same library type. A poly(A) batch and a ribodepleted batch are not one experiment with a missing covariate. They are two measurements.

Deposition records in the European Nucleotide Archive show how public datasets label library selection. Read those fields when you reuse someone else's counts. "RNA-seq" alone does not say which molecules were invited. Method sketches on protocols.io are useful for seeing workflow shape. The probe set and the selection field in your own record are the facts that matter.

Research use and sample governance

Library type is a research design choice. It does not authorise a diagnostic report, and it does not set a biosafety level. Human tissue, infected cultures and environmental microbes stay under the institutional rules that govern the source material. Depletion and fragmentation reagents have their own hazard notes. Follow those notes for the kit class you handle.

A warm delay pushes the library choice

In a hot, humid laboratory, the minutes between excision and stabilisation are long enough for tails to shorten and for RNase to open transcripts. A power cut that thaws a shipment does the same thing in the freezer log. If an integrity trace then shows the RNA is broken, poly(A) selection will narrate the damage. Ribodepletion is often the fairer library for that material, and the damage still belongs in the methods. Record the delay, the trace and the selection together so the differential table is not asked to explain a warm afternoon.

What the enquiry needs

State the organism, whether the RNA is intact, the selection you think you need, whether the library must be stranded, the number of biological replicates, and the contrast. Reagent and instrument classes for sequencing sit in the genomics and sequencing catalogue. The sample-to-result path is the nucleic acid analysis pathway.

Use the mRNA sequencing enquiry reference and the differential expression analysis enquiry reference as enquiry references when you want those methods discussed. Send the scientific requirement through the quote request and ask whether a quotation is possible. A library type can be discussed there. The pages are how you specify the work, with the selection choice written in your own words.