A well-designed GLORI-seq project starts with a clear biological comparison. Unlike antibody-based m6A methods that report enrichment peaks, GLORI-seq provides absolute stoichiometry at single-base resolution — the fraction of transcripts carrying m6A at each site. This quantitative output means the experimental design must support a statistical comparison, not just a catalog of sites. If you have previously evaluated antibody-free detection methods, our overview of DART-seq and antibody-free m6A detection describes how GLORI-seq differs from enzymatic alternatives.

This guide walks through the project design decisions that determine whether a GLORI-seq experiment produces interpretable differential m6A results: defining comparison groups, choosing the number of replicates, selecting appropriate controls, deciding whether to integrate RNA-seq, and preparing the information a service provider needs to generate an accurate quote.

Figure 1: A GLORI-seq project design flows from a defined biological comparison through replicate planning, control inclusion, and quantitative analysis.

What GLORI-seq Measures — And Why Design Matters

GLORI-seq uses a chemical reaction — glyoxal and nitrite — to deaminate unmethylated adenosines to inosines, which are read as guanosines during sequencing. m6A residues are protected from deamination by the methyl group and remain read as adenosine. The fraction of reads retaining A at each site directly reports the m6A modification level at that position.

This is fundamentally different from antibody-based enrichment methods such as MeRIP-seq, where the output is a peak — a region enriched in the IP relative to input — that reflects relative m6A abundance rather than absolute stoichiometry. In GLORI-seq, the output for each site is a number between 0 and 1: the proportion of transcripts modified.

Because the readout is quantitative, the design must be quantitative too. A poorly designed comparison — too few replicates, unmatched conditions, no spike-in calibration — produces modification fractions that cannot be interpreted statistically. A well-designed comparison produces a list of sites where the modification level differs between conditions, with confidence estimates attached.

For the underlying chemistry and step-by-step workflow, see our guide to GLORI-seq principles and experimental steps. The rest of this article assumes familiarity with the basic mechanism and focuses on project design.

Defining Your Comparison — Groups, Conditions, and Questions

The most important design decision is the comparison structure. Every GLORI-seq project starts with a question that can be expressed as a contrast between groups.

Two-group comparison. This is the simplest and most common design: treatment versus control, knockout versus wild-type, or disease versus healthy. With GLORI-seq, the comparison is made at each m6A site — “does the modification fraction differ between conditions at this position?” A typical two-group design uses 2–3 biological replicates per group, consistent with published GLORI-seq studies that have examined m6A dynamics under heat shock, hypoxia, and writer knockdown conditions.

Multi-condition design. When the biological question involves more than two conditions — for example, a time course, a dose response, or multiple genotypes — the design needs enough replicates per condition to support the planned contrasts. For a three-condition experiment with pairwise comparisons, plan at least 2–3 replicates per condition. For a time course, consider whether the analysis will treat time as a continuous variable or as discrete groups, since this affects both replicate allocation and the downstream statistical model.

Paired or matched design. If samples come from the same individual before and after treatment, or from matched tissue pairs, the pairing should be declared in the design. Paired analysis can substantially increase power for detecting condition-dependent m6A changes because it controls for between-individual variability in baseline modification levels.

Factorial design. When two or more factors interact — for example, genotype and treatment — the design must include all factor combinations with replicates in each cell. A 2×2 factorial design with 2 replicates per cell requires 8 samples total. Factorial designs are powerful but demand more sequencing resources; confirm that the interaction term is biologically meaningful before committing the additional samples.

One practical step before finalizing groups: write a one-sentence statistical contrast for each comparison you intend to make. If you cannot express the contrast clearly, the group structure probably needs refinement.

Replicates and Statistical Considerations

Biological replicates are non-negotiable in quantitative m6A studies. Technical replicates — resequencing the same library — do not substitute for biological replicates, which capture the natural variation in modification levels between independent samples.

How many replicates? Published GLORI-seq studies typically use 2–3 biological replicates per condition. Two replicates is the minimum; three is safer, particularly when effect sizes are expected to be modest or when sample-to-sample variability is unknown. With only two replicates per group, a single outlier replicate can mask or exaggerate a differential signal. Adding a third replicate per group is often a better use of budget than deeper sequencing of two replicates.

What statistical framework applies? Because GLORI-seq produces a modification fraction at each site — essentially a proportion — differential analysis is typically performed with models designed for count or proportion data. The GLORI-tools pipeline provides site-level modification fractions. Downstream differential testing can use approaches similar to those applied in differential DNA methylation analysis, where beta-binomial models or logistic regression frameworks account for both the proportion and the read depth supporting each site.

Sequencing depth. The original GLORI-seq publication identified approximately 80,000 m6A sites with 50 Gb of sequencing per sample in mammalian cells. Deeper sequencing (approximately 140× coverage) increased detection to over 210,000 sites. For differential analysis, higher coverage improves the precision of modification fraction estimates, particularly at sites with low or moderate modification levels. If budget is constrained, prioritize replicate number over per-sample depth: consistent detection across replicates is more valuable for differential testing than deep coverage in fewer samples.

Controls and Calibration

GLORI-seq includes built-in controls that differ from those needed in antibody-based methods, but additional calibration measures improve data quality.

Untreated RNA control. An aliquot of each sample should be reserved before the chemical deamination step and sequenced separately. This untreated control allows estimation of background — reads that appear to retain A at positions that are not actually m6A. In practice, the GLORI-seq protocol reports that in vitro transcribed RNA (which carries no m6A) yields only approximately 1,856 false-positive sites, with 0.09% overlap with genuine m6A positions.

Spike-in standards. Synthetic RNA standards with known m6A stoichiometry — for example, oligonucleotides methylated at defined fractions — can be added to each sample before the deamination reaction. The conversion efficiency measured on these standards provides a per-sample calibration factor. If the standard with 50% methylation reads out at 48% after GLORI-seq processing, the 2% offset can be applied to all sites in that sample.

Conversion efficiency QC. The deamination efficiency — the fraction of unmethylated A converted to I — should be monitored per sample. The protocol reports approximately 99% conversion of unmethylated A, with less than 4% C-to-U and approximately 3% G-to-X side reactions. Samples with conversion efficiency below a laboratory-established threshold (typically 95% or higher) should be flagged for re-processing or exclusion.

Batch and lane effects. If samples are processed in multiple batches or sequenced across multiple lanes, randomize condition assignments across batches rather than processing all samples from one condition together. Batch-confounded designs cannot distinguish treatment effects from processing effects.

Figure 2: Three layers of controls — untreated RNA, spike-in standards, and conversion efficiency monitoring — support reliable differential m6A quantification.

Integrating RNA-seq with GLORI-seq

A common question in GLORI-seq project design is whether to include matched RNA-seq data. The answer depends on what claim the project needs to support.

When RNA-seq adds value. If the biological question involves distinguishing changes in m6A modification from changes in transcript abundance, RNA-seq is essential. A site that shows increased m6A fraction in treatment versus control could reflect either higher methylation of the transcript or higher expression of the transcript — both increase the absolute number of modified reads. RNA-seq expression data allows the two effects to be separated. For a detailed discussion of combined analysis strategies, see Integrating RNA-seq and Epigenomic Data Analysis.

When RNA-seq can be deferred. If the primary goal is to identify candidate sites with large modification differences — and expression-level confounds are acceptable at the discovery stage — GLORI-seq alone is sufficient. Many projects use GLORI-seq for discovery and add RNA-seq only for the subset of samples that carry the strongest candidate sites, reducing total sequencing cost.

Practical integration. When RNA-seq is included, it should be performed on an aliquot of the same RNA used for GLORI-seq, not on a separate RNA extraction. This ensures that expression and modification measurements are made on comparable material. Standard RNA-seq depth — 30–50 million paired-end reads per sample — is typically sufficient for expression normalization.

What to Prepare Before a Project Consultation

A GLORI-seq project consultation moves faster when the following information is ready. Each item directly affects the experimental design, the quote, or both.

Biological question and comparison structure. Write a clear statement of the comparison: “We are comparing m6A modification levels between wild-type and METTL3-knockout HEK293T cells, with three biological replicates per group.” If the comparison involves more than two conditions, describe all groups and the contrasts of interest.

Sample information. For each sample, provide the species, tissue or cell type, expected RNA yield, and RNA quality metrics if available (RIN or RQN). GLORI-seq requires microgram-level input — the protocol uses 1 μg of poly(A)-selected RNA as the standard starting amount. Low-input variants are under development but are not yet standardized; if RNA is limiting, discuss feasibility with the provider before finalizing the design.

Replicate count and rationale. State how many biological replicates per condition and why. If the number is constrained by sample availability, the provider can advise on whether the design can still support the planned statistical comparisons.

Whether RNA-seq is needed. Indicate whether matched RNA-seq is part of the plan, and if so, confirm that aliquots of the same RNA extraction will be used for both assays.

Reference genome and annotation. Specify the genome build and transcript annotation version. This matters because m6A site calling depends on accurate alignment and transcript annotation, particularly for assigning sites to gene features (5′ UTR, CDS, 3′ UTR).

Preferred analysis outputs. Describe what outputs the project requires — a site-level modification table, differential m6A results with effect sizes and confidence intervals, annotation of differentially modified sites by gene feature, and enrichment analysis of affected pathways or sequence motifs.

For projects where the experimental design is still taking shape, or where multiple assay options are under consideration, RNA modification services at CD Genomics support method selection and study design across the full epitranscriptomics workflow, from discovery through validation.

Figure 3: Preparing these six items before a project consultation ensures the provider can assess feasibility and generate an accurate quote.

Summary

A GLORI-seq project designed for differential m6A quantification needs more than an assay choice — it needs a comparison structure, a replicate plan, appropriate controls, and a decision about RNA-seq integration.

The comparison structure determines what statistical contrasts can be tested. Biological replicates determine how reliably those contrasts can be made. Spike-in controls and untreated RNA references determine whether the quantification is calibrated. RNA-seq determines whether modification changes can be separated from expression changes.

Much of the planning is front-loaded — clarifying the question, defining groups, choosing controls — but that investment is what separates a project that produces a list of candidates from one that produces interpretable, publication-ready differential m6A results. For assistance designing a GLORI-seq study or selecting the most appropriate m6A method for your project, Epigenomic Data Analysis services offer integrated support from experimental design through data interpretation.

FAQ

1. How many biological replicates do I need for a GLORI-seq differential m6A study?

A minimum of 2 biological replicates per condition is standard in published GLORI-seq studies, but 3 replicates per condition is recommended — especially when effect sizes are expected to be modest or when baseline variability between samples is unknown. Adding a third replicate per group is typically a better investment than deeper sequencing of two replicates.

2. Do I need matched RNA-seq data for every GLORI-seq project?

No. RNA-seq is essential when the project needs to distinguish m6A modification changes from transcript expression changes. If the primary goal is discovery of sites with large modification differences, GLORI-seq alone is sufficient. RNA-seq can be added later for the most promising candidates.

3. What spike-in controls are recommended for GLORI-seq calibration?

Synthetic RNA standards with known m6A stoichiometry — oligonucleotides methylated at defined fractions such as 0%, 25%, 50%, and 100% — can be spiked into each sample before the deamination reaction. The measured versus expected modification fractions on these standards provide per-sample calibration factors.

4. How much RNA and sequencing depth does a GLORI-seq project require?

The standard protocol uses approximately 1 μg of poly(A)-selected RNA per sample. Recommended sequencing depth is approximately 50 Gb per sample for comprehensive m6A site detection in mammalian transcriptomes, with deeper sequencing (140× coverage or more) improving sensitivity at lowly modified sites.

5. Can GLORI-seq be used with FFPE or degraded RNA samples?

GLORI-seq has not been extensively validated on FFPE or highly degraded RNA. The chemical deamination protocol requires intact poly(A)-selected RNA. For degraded or low-quality RNA samples, consult the service provider about feasibility before committing to the project design.

References

1. Shen, Weiguo, Hanxiao Sun, Cong Liu, et al. "GLORI for absolute quantification of transcriptome-wide m6A at single-base resolution." Nature Protocols, vol. 19, 2024, pp. 1252–1287. DOI: 10.1038/s41596-023-00937-1
2. Liu, Cong, Hanxiao Sun, Yunpeng Yi, Weiguo Shen, Kai Li, Ye Xiao, Fei Li, et al. "Absolute quantification of single-base m6A methylation in the mammalian transcriptome using GLORI." Nature Biotechnology, vol. 41, 2023, pp. 355–366. DOI: 10.1038/s41587-022-01487-9
3. Dominissini, Dan, et al. "Topology of the human and mouse m6A RNA methylomes revealed by m6A-seq." Nature, vol. 485, 2012, pp. 201–206. DOI: 10.1038/nature11112

CD Genomics provides epitranscriptomics services for research use only and they are not intended for clinical diagnosis or treatment.