RNA sequencing measures gene expression across the whole transcriptome simultaneously — capturing which genes are turned on or off, how strongly, and in response to what. The standard method for differential expression analysis, pathway studies, and transcript discovery.
Every cell in an organism carries essentially the same genome. What differentiates a neuron from a liver cell, a pathogen in exponential growth from one under antibiotic stress, or a tumour from the surrounding tissue, is not primarily the DNA sequence — it is the pattern of gene expression. RNA sequencing measures this expression landscape across the entire transcriptome in a single experiment, capturing the quantity and identity of every messenger RNA present in a sample at the moment of collection.
The method has displaced microarrays as the standard transcriptomics approach, for reasons that go beyond simple technical improvement. RNA-seq is not limited to pre-defined probe sets: it captures everything — including novel transcripts, alternative splicing events, non-coding RNAs, and fusion transcripts that a microarray would never detect. It also generates count data with a broad dynamic range, allowing detection of both highly abundant transcripts and rare ones in the same experiment. The current challenge in RNA-seq is not technical sensitivity but experimental design: replication, batch effects, and the choice of library preparation strategy have far more impact on study quality than sequencing depth, and they are harder to fix after the fact.
Library preparation strategy is the decision that most commonly gets overlooked. For standard eukaryotic RNA-seq — well-preserved RNA from tissue or cells, where ribosomal RNA has already been removed by poly-A selection — the choice is straightforward. Poly-A selection is fast, cost-efficient, and well-supported by bioinformatics tools. The moment your RNA is degraded (FFPE tissue, clinical biopsies, aged samples), non-polyadenylated (bacterial RNA, many lncRNAs), or from an organism without a well-annotated poly-A signal, rRNA depletion becomes the appropriate approach. It adds cost and complexity, but it is the only route to reliable results from those sample types. We’ll discuss this with you before any library preparation begins.
How much sequencing depth you need depends on your question. Differential expression analysis with a well-annotated reference genome and a binary treatment comparison typically works well at 20–30 million reads per sample. Splicing and isoform analysis requires 50–100 million. Novel transcript discovery — particularly in organisms with incomplete annotations — benefits from higher depth still, and from longer reads. We will always help you determine the right depth for your specific question during the consultation; sequencing less than your study needs is a waste of money that’s expensive to correct.
Three biological replicates is the absolute minimum for differential expression analysis; four to six are strongly recommended for publication-quality results, especially in complex systems or when effect sizes are expected to be modest. Technical replication (running the same sample twice) adds almost nothing to statistical power. Pooling RNA from multiple individuals into a single sample eliminates your ability to detect biological variability — which is usually the entire point. These are not preferences — they are practical requirements for publishable results, and we’ll always ask about your replication strategy before agreeing a project scope.
NA-seq requires well-preserved RNA — degraded input is the most common cause of failed experiments. If your samples are FFPE, clinical biopsies, or otherwise at risk of RNA degradation, ask us for advise on sample handling.
Our free consultation is made for this
Free consultation → →The core application: comparing transcriptomes between treatment and control, knockout and wild-type, disease and healthy tissue, different time points, or different growth conditions. RNA-seq provides unbiased, whole-transcriptome measurement with count-based statistical models (DESeq2, edgeR) that have well-understood properties. Any biological question that reduces to "which genes are more or less active under condition X versus Y" is a differential expression problem, and RNA-seq is the right tool.
Understanding how bacteria respond to antibiotics at the transcript level — which resistance mechanisms are induced, which stress response pathways are activated, which metabolic shifts occur — informs drug combination strategies, mechanisms of tolerance, and the discovery of novel targets. Bacterial RNA-seq requires rRNA depletion (most bacterial RNA is ribosomal) and different library prep strategies than eukaryotic RNA-seq. We have validated workflows for the organisms most commonly studied in clinical and environmental AMR contexts.
Formalin-fixed paraffin-embedded tissue is the most common archival material in clinical research — but formaldehyde cross-linking degrades RNA and introduces noise. With rRNA depletion libraries optimised for low-integrity RNA (RIN 2–5), and appropriate bioinformatics filtering, RNA-seq from FFPE is now a feasible and well-validated approach. It won't match the quality of fresh-frozen tissue, and it requires more careful QC, but it opens archived clinical cohorts — including retrospective cancer studies — to transcriptomic analysis that would otherwise be impossible.
For organisms with incomplete or poorly annotated genomes, or any experiment where the goal is to discover previously unannotated transcripts, RNA-seq provides an unbiased scan of the transcribed genome. Unlike microarrays, RNA-seq is not limited to known sequences — it will detect any transcript present in your sample, including antisense transcripts, small RNAs, novel splice variants, and fusion genes. Higher depth (50–100M reads) and longer read lengths improve sensitivity for rare and novel transcripts.
Profiling both host and microbial transcriptomes simultaneously from the same infected sample. Typically used for in vitro infection models, organoid studies, or animal models where tight experimental control is possible. Requires careful design — particularly around the proportion of microbial RNA in the total sample — and appropriate depth to achieve statistical resolution in both compartments. We design the experimental approach alongside you before any cells or animals are committed.
Specifications for standard bulk RNA-seq projects. Speak to us about single-cell, long-read, or dual RNA-seq.
Five stages from first contact to data delivery. RNA quality determines everything downstream — which is why we assess it before any library preparation begins, not after.
We discuss your experimental question, organism, sample types, expected number of conditions and replicates, and any RNA integrity concerns. We'll confirm library preparation strategy (poly-A versus rRNA depletion), sequencing depth, and whether any add-on bioinformatics are appropriate for your analysis plan. If your study design has statistical issues — too few replicates, confounded variables, inappropriate pooling — we'll raise them here, not after the sequencing is done.
Always free. Study design problems are much cheaper to fix before sequencing than after.
Written proposal covering library prep strategy, depth, add-on analyses, expected turnaround, and pricing. All add-ons (differential expression, pathway analysis, isoform analysis) are specified here so the analysis plan is aligned with the sequencing strategy from the start.
Nothing proceeds without written sign-off.
Ship total RNA frozen on dry ice, or cells / tissue per our guidelines for extraction at our facility. On receipt, we assess RNA integrity (TapeStation RIN) and quantify by Qubit. Samples below minimum thresholds are flagged before library preparation — not after. For FFPE samples, we assess DV200 (the fraction of RNA fragments >200 nt) as the relevant metric.
Contact us before shipping clinical or regulated biological material. Cold-chain documentation requirements vary.
Delivery of FASTQ files via secure link.
Three biological replicates is the minimum below which most statistical models for differential expression analysis produce unreliable results. Published best-practice guidelines (ENCODE, MAQC) and simulation studies consistently show that moving from 3 to 4 or 5 replicates increases statistical power more than doubling sequencing depth at 3 replicates. This matters most when effect sizes are expected to be modest, when there is high biological variability between samples, or when you are studying complex phenotypes. If your experimental constraint is cost, reducing sequencing depth slightly to fund a fourth replicate is almost always the right trade-off. Technical replicates — running the same RNA library twice — add very little in modern RNA-seq experiments and should not be confused with biological replicates. Pooling multiple biological samples into one replicate eliminates your ability to measure variability and should generally be avoided unless population-level averages are your explicit study goal.
It depends on why the RNA is degraded and how degraded it is. FFPE tissue routinely yields RNA with RIN 2–4, and there are validated workflows for generating RNA-seq data from these samples using rRNA depletion libraries and FFPE-specific protocols. The relevant metric for FFPE is DV200 (the percentage of RNA fragments >200 nucleotides), not RIN. Results from FFPE RNA-seq are noisier than from fresh-frozen material — expect more variability and lower sensitivity for low-abundance transcripts — but the data is usable for differential expression analysis, particularly for highly expressed genes. For degraded RNA from sources other than FFPE — badly stored samples, samples without stabilisation, repeated freeze-thaw cycles — recovery is less predictable. Share your DV200 or Bioanalyzer trace with us before committing any samples; we’ll give you an honest assessment of what’s recoverable.
Answer: Poly-A selection enriches for messenger RNAs by capturing their poly-A tail — it’s cost-efficient, well-characterised, and the default for standard eukaryotic transcriptomics from high-quality RNA. Its limitations: it only works if your RNA is intact (degraded RNA lacks poly-A tails), it misses non-polyadenylated transcripts (including most bacterial RNA, many lncRNAs, histone mRNAs), and it requires a minimum RNA integrity. rRNA depletion removes ribosomal RNA by hybridisation and leaves everything else — including degraded mRNA, non-polyadenylated transcripts, and regulatory RNAs. It’s more expensive and produces more background, but it works on FFPE RNA, bacteria, organisms with non-standard poly-A signals, and any application where the transcriptome is more than just mRNA. If you’re unsure, tell us about your organism, sample type, and preservation method — the choice is usually clear once we know the context.
ell us about your organism, sample type, experimental conditions, and approximate number of replicates. We'll respond as fast as we can — or arrange a call if the study design needs more discussion first.