RNA-seq
Oligo-dT beads capture polyadenylated mRNA from total RNA before fragmentation and cDNA synthesis. Reads are mostly exonic and the library is efficient, but it needs intact RNA (RIN ≥ 7–8) and misses non-polyadenylated transcripts.
Our ruleSimulations prepare you for the bench but don't replace hands-on practice.
Poly(A) selection
Oligo-dT beads capture polyadenylated mRNA from total RNA before fragmentation and cDNA synthesis. Reads are mostly exonic and the library is efficient, but it needs intact RNA (RIN ≥ 7–8) and misses non-polyadenylated transcripts.
How it differs from the standard method
- Degraded RNA gives 3′-biased coverage and length-dependent loss of counts.
- Histone mRNAs and many non-coding RNAs, which lack poly(A) tails, are lost.
- Two rounds of capture keep residual rRNA to a few percent.
1Counting transcripts by sequencing them
RNA-seq turns the RNA of a sample into a library of short cDNA fragments, sequences millions of them and assigns each read to the gene it came from. The number of reads a gene collects is proportional to how many of its transcripts were in the sample and to how long they are, so after correcting for the size of each library the counts compare one sample with another.
Capture mRNA by its tail
Oligo-dT beads pull out polyadenylated RNA. Clean, mostly exonic reads, but only for intact RNA: a broken transcript loses everything upstream of the break.
Remove the ribosomes instead
Probes hybridise to ribosomal RNA, which is then removed. Keeps non-polyadenylated and fragmented RNA, so it works on degraded samples, with many intronic reads from pre-mRNA.
Ligate to 18–30 nt RNAs
Adapters are ligated straight onto miRNAs and other small RNAs, and the library is size-selected away from adapter dimers.
2Where the power to find a change comes from
A gene's count varies between replicates for two reasons. Counting noise shrinks as you sequence deeper.
3How it fails
Degraded RNA with poly(A) capture gives coverage piled up at the 3′ end of genes, and long transcripts lose counts in proportion to how degraded each sample is. Genomic DNA in a random-primed library spreads reads into introns and between genes.
- Small RNA pitfalls
- Adapter dimers when adapters are in excess over RNA ends; loss of miRNAs in column kits that discard RNA under ~200 nt; ligation bias that distorts the relative levels of different miRNAs.
Key terms
Practise these as flashcards →An independent sample of the biology (another animal, donor or culture), as opposed to sequencing the same RNA twice.
The extra-Poisson variability of a gene's counts between biological replicates; its square root is the biological coefficient of variation.
Counts per million reads, and transcripts per million, which also divides by transcript length so genes within a sample can be compared.
A per-sample scaling that puts libraries of different depth and composition on a common scale before testing.
The base-2 logarithm of the ratio of normalised expression between conditions: 1 means doubled, −1 halved.
The expected share of false positives among the genes called; 5% is the usual threshold.
RNA integrity number (1–10, from the electrophoretic trace) and the percentage of RNA fragments above 200 nt, the metric used for FFPE samples.
The share of reads that are copies of a molecule already read; high values mean the library had too few distinct molecules for the depth.
A library that keeps the information about which DNA strand a transcript came from, usually by marking the second cDNA strand with dUTP.
When to use it
- Finding which genes change between conditions, genotypes or treatments, without choosing the genes in advance.
- Profiling archived FFPE tissue or partly degraded RNA, using rRNA depletion rather than poly(A) capture.
- Discovering isoforms, fusions, non-coding RNAs or allele-specific expression from the same data.
- Profiling miRNAs and other small RNAs with a dedicated small RNA library.
Compared with alternatives
Faster and cheaper for a handful of known genes; the usual way to validate a few RNA-seq hits in new samples.
Cheaper for very large cohorts on a well-annotated organism, but limited to the probes on the chip and to a narrower dynamic range.
Prefer it when the question is which cells change, or when a bulk change could come from a shift in cell-type proportions.
Long-read, direct RNA and ribosome profiling answer isoform, modification and translation questions that short-read counts cannot.
Covers the alignment, counting and quality-control steps that turn the reads into the count table tested here.
Sub-techniques
Choosing one here switches the whole page, including the protocol and the lab, to it.
Oligo-dT beads capture polyadenylated mRNA from total RNA before fragmentation and cDNA synthesis. Reads are mostly exonic and the library is efficient, but it needs intact RNA (RIN ≥ 7–8) and misses non-polyadenylated transcripts.
- Degraded RNA gives 3′-biased coverage and length-dependent loss of counts.
- Histone mRNAs and many non-coding RNAs, which lack poly(A) tails, are lost.
- Two rounds of capture keep residual rRNA to a few percent.