Questions before an RNA-seq enquiry

The expensive mistake in an RNA-seq project is often the first message, the one that says only "please sequence these tubes". The library chemistry, the reference genome and the contrast are still free at that moment. They become structural once the reads exist. This troubleshooting note is the list of questions that keep a brief from producing the wrong files. How expression studies are shaped overall is in from cells to a gene expression result. Nothing here is a price list or a booking.

A thin brief is the failure

A useful enquiry is a scientific specification. It can be short. It cannot be vague about the organism, the RNA fraction, or the sentence you hope to write. Troubleshooting a study that has already been sequenced is mostly archaeology. Troubleshooting the brief is cheaper because the branch is still open: measure integrity, change the selection method, add a replicate, or decide the question is really a qPCR question.

Two public browsers, Ensembl and the UCSC Genome Browser, are where you confirm that the organism and the gene models exist before you name them in the brief. A short-read orientation from Illumina is background for the instrument class, not a form you fill with defaults.

Each question prevents a different wrong library

Organism and sample type. Species, tissue or cell line, and whether the material is fresh, frozen or already extracted. The reference genome follows the organism. A close species is not a harmless substitute. State the assembly or annotation release if you depend on particular gene models. If several organisms are in the tube, say so. A single reference will mis-assign those reads.

Poly(A) selection or ribosomal depletion. Poly(A) capture follows tailed eukaryotic RNA and is a poor match to many bacterial transcriptomes and to heavily degraded RNA. Ribosomal depletion keeps a broader fraction and leaves a ribosomal leftover you should still check. The wrong choice cannot be edited out of the FASTQ file. If you do not know which fraction your question needs, that ignorance is the thing to discuss, not a default.

Stranded or unstranded. If antisense or overlapping genes matter, ask for a stranded class and say which orientation the kit uses if you know it. If they do not, an unstranded library may be enough, and the analyst should be told that explicitly. "Standard RNA-seq" does not answer the question.

Read length and pairing. Paired reads help with splicing and with stranded assignment. A short single read can be enough for a simple gene count and weak for isoforms. Name the length and the pairing you have in mind, and why. Do not copy a configuration from a paper whose fragment sizes differ from yours.

Depth intention. Say what the study has to see. Common messenger RNA with a clear contrast is a different depth problem from rare splice forms or from a library still full of ribosomal RNA. Ask the discussion to turn that intention into a plan. Do not paste a universal read count as though every organism and every library class shared one number.

Replicates and the contrast. Name the biological unit, the groups, and what will be compared. A pile of tubes without a contrast cannot be turned into a differential expression result by analysis. If the unit count is still one per group, say so. The honest deliverable may be a pilot quantification rather than a treatment claim.

Integrity, if you know it. An RNA integrity number or an equivalent score, such as a fragmentation metric used for degraded material, tells the chemistry conversation whether poly(A) selection is even plausible. If you have not measured it, write unknown. Absorbance ratios are not a substitute. They do not prove the RNA is intact.

Analysis scope. QC only, alignment and a count matrix, a differential table, or figures as well. These are different pieces of work. A FASTQ delivery disappoints a group that wanted a contrast. A contrast disappoints a group that wanted to run its own model. Name the scope, the reference, and the contrast in the same paragraph.

Reference. Genome or transcriptome, release, and whether you need a custom annotation. Counts are features in that reference. A later argument about a missing gene is often an argument about the wrong release.

Reference genes in the qPCR sense are not an RNA-seq input setting. If the enquiry is really about three named transcripts and a normaliser, say that. The better method may be RT-qPCR, described in RT-qPCR for relative expression, and the brief should change rather than force a sequencer to answer a primer question.

Facts to gather, not a price list

You do not need a commercial term to write this brief, and this page will not invent one. You do need the scientific facts above, plus practical constraints that change the method: total RNA already in a buffer, a species without a public genome, or a sample that cannot leave a certain biosafety setting. Those constraints belong in the message.

How a laboratory writes a sourcing note in general is in how to write a laboratory sourcing enquiry. Sequencing-specific checks that happen once a run is real are in what to check before a sequencing run and in commissioning a sequencing or proteomics study. Use them as sister lists. Do not skip the chemistry questions because a commissioning checklist looks complete.

Branches when the answer is still unknown

Unknown organism details: stop and identify the material. A library without a reference can sometimes be assembled, and that is a different project. Do not start it by accident.

Unknown integrity: measure it, or agree that fitness to assay is the first question. Do not select poly(A) on a hunch about a tissue that sat warm.

Unknown strandedness requirement: look at one locus you care about on a browser. If an antisense or an overlap would change the interpretation, choose stranded and say so. If you truly do not care, say unstranded is acceptable.

Unknown depth: describe the rarest feature you need. Let the method discussion respond. Inventing a precise read count to look prepared is how incompatible studies get the same number written on them.

Unknown analysis: decide whether a count matrix would already answer you. If a statistician in the group will fit their own model, ask for counts and a sample sheet, and say which identifier joins them. If you need a differential table, name the contrast so the sheet can be checked against it.

A failed control at the RNA stage branches backward. If integrity is poor because of extraction, another library kit will not repair the sample. Protecting RNA during extraction is the upstream note.

Question, and what goes wrong if you skip it

Briefs that produce the wrong files

A brief that names a kit brand and nothing else will be answered with that kit's defaults, which may be unstranded, poly(A), and a depth that suited a different organism. A brief that asks for "publication-quality analysis" without a contrast invites a heatmap of a batch effect. A brief that hides poor RNA because the collection was difficult will buy reads of whatever fragments remain.

Write the limitation in the message. Degraded, unknown integrity, one replicate, no reference genome: each of those has a method response. Hidden, they become a surprise in the QC report, which is the wrong time to choose a different assay.

Research questions only

An enquiry is not a diagnostic order and not a biosafety approval. Say what the material is so the receiving conversation can follow institutional rules for human or infectious samples. Do not send identifiers built from patient names. Do not expect a research count to come back as a clinical interpretation. Chemical handling of library reagents stays with the people who run the method, under their own assessment.

Integrity before a warm journey

If RNA will travel, the integrity you quote should be the integrity you expect on arrival, or you should plan to repeat the measurement after the handoff. Heat in transit rewrites a score that was true at the bench. A power cut in a shipper is a different sample. Say how the tube will be packed and what should happen if it arrives warm. The brief can include that branch: do not make a library from a tube that warmed, tell us instead.

Humidity and delays matter more to RNA in a half-open tube than to a sealed frozen aliquot. Close the specification with who is responsible for the freeze before the journey. A discussion that assumes the sample is intact will follow whatever you wrote. If you wrote nothing, it will assume.

How to open the discussion

Use the questions on this page as the body of the message. The mRNA sequencing enquiry reference is where library method can be discussed. The differential expression analysis enquiry reference is where scope, reference and contrast can be discussed. Read both as prompts. Ask which method class fits the brief. Do not treat either page as evidence that a sequencer is already allocated to the tubes.

Consumables related to sequencing are grouped in the genomics and sequencing catalogue. The surrounding path is nucleic acid analysis. Send the filled questions through the quote request.