Reading a modification ratio
A ratio at a position comes from the composition of reads at that position, and it describes the share of molecules that carry the mark there. Read as a per-site quantity, it is the strongest statement a conversion-based method makes.
01What this article answers
At a site called by a conversion-based method, the readout is a ratio built from the reads at that single position. In m6A work, unmodified adenosine is converted and copied as guanosine while m6A is copied as adenosine, so the balance between the two read types reports the fraction of the sample’s molecules that carry the modification at that position.
That number is a property of one position, not of the transcript. Two sites on the same molecule can carry very different ratios, and a single value does not describe the transcript as a whole.
The sections below set out where the ratio comes from, how to compare it between samples, and what it does not say.
02Where the ratio comes from
The value is produced by the same signal that locates the site, which is what makes position and level arrive together. Different methods build it from different measurements.
| Method | What produces the level at a position |
|---|---|
| m6A Modification Seq (CAM-seq) | The A-to-G balance at the site, from cooperative-catalysis deamination of unmodified adenosine |
| m6A Modification Seq (GLORI2-seq) | The converted-to-unconverted ratio at the position, reported as absolute stoichiometry |
| m1A Modification Seq (m1A-Quant-seq) | The mutation rate at the base, read against a demethylated background control |
| Pseudouridine (Ψ) Seq (BACS-seq) | The protected fraction of molecules at that position, reported on an absolute scale |
The pattern is the same across them: the level is read out of the composition of reads at one position rather than inferred from how much material a capture step collected. That is why a ratio from these methods is not an enrichment score and cannot be compared with one.
03How to compare ratios between samples
A ratio supports a comparison when the two sides are read the same way. Three rules follow from that.
- Compare the same position. A difference between conditions is a change at one site, so the site is the unit of comparison rather than the transcript.
- Keep the site set fixed across the groups. Comparing one set of sites in one group with a different set in another mixes two variables into a single number.
- Use an absolute-scale readout where transcriptome-wide levels are being compared. Where the aim is to compare levels between sites and between samples, the readouts described as absolute are the ones built for it.
- Read the ratio next to the abundance data. Transcript levels and modification levels can move in the same experiment, so a change in a ratio is read alongside expression rather than instead of it.
04Limits and common misreadings
- A ratio is not a functional effect. It reports how much of a position carries the mark, which is a statement about the RNA rather than about what the mark does.
- It is not an average over the transcript. Averaging site-level values into one transcript-level figure describes a quantity the assay did not measure.
- A site without a call has no ratio. The level rests on the site being resolved in the first place, so the site table and the ratio are read together.
- Ratios from different readout types do not mix. A modification ratio and an enrichment score come from different measurements, even when both describe the same mark.
06FAQ
Does a modification ratio describe a transcript or a position?
A position. The number is read from the composition of reads at that single nucleotide, so a transcript carries as many ratios as it has called sites. Two sites on the same molecule can carry very different ratios, and averaging them into one transcript-level figure describes a quantity the assay did not measure.
Why is the ratio not reported as an enrichment score?
Because nothing is enriched. The value comes from the composition of reads at the position after a defined conversion, which is a different measurement from capture-based enrichment. A modification ratio and an enrichment score therefore do not mix, even when both describe the same mark.
Can ratios be compared between two experiments?
They can be compared site by site when both experiments are read the same way and the same positions are used, since a difference between conditions is a change at one site and the site is the unit of comparison. Comparing one set of sites against another mixes two variables into a single number.
How should a site with a very low ratio be read?
As the level reported for that position, and alongside the site call itself, because the ratio rests on the site being resolved in the first place. Whether such a site matters to a study is a biological question rather than a property of the ratio.