eQTL mapping in single cells: New data, new models

eQTL mapping in single cells: New data, new models

by International Statistical Genetics Workshop

Formal & Physical Sciences Biology PSALife sciences: general issuesPSAKGenetics (non-medical)
🎙 International Statistical Genetics Workshop 👥 3K 📅 May 18, 2026 ⏱ 12 min 👁 241 📄 lecture 🧭 2026-08-16
Available in: English (current) Français

Keywords

eQTLsingle-cell RNA-seqpseudobulkPoisson regressionmixed model

Summary

This lecture introduces statistical methods for mapping expression quantitative trait loci (eQTL) in single-cell RNA-seq data. It contrasts bulk and single-cell approaches, highlighting the ability to detect cell-type-specific effects. The pseudobulk method aggregates cells per donor and cell type, but it loses information about cell-level distribution, covariates, and cell number imbalance. To address these issues, the lecture proposes Poisson regression for count data, with a log link and library size offset. However, naive Poisson regression ignores correlation among cells from the same donor, leading to inflated false positives. The solution is a Poisson mixed model with a donor-specific random effect to account for within-donor correlation. The lecture then introduces SAIGE-QTL, a scalable framework that combines these elements, also supporting context-dependent and rare variant tests. A concrete result from the oneK1K cohort shows SAIGE-QTL identifies 48% more eGenes than pseudobulk, demonstrating the power of modeling single cells directly.

147 words

Critical Evaluation

Value of the Information & Strength of the Argument

The lecture provides a clear and logical progression from the limitations of bulk eQTL mapping to the statistical challenges of single-cell data, and then to the solutions embodied in SAIGE-QTL. The argumentation is solid, building on established statistical principles (Poisson regression, mixed models) and illustrating with a concrete example. The value lies in its pedagogical clarity and the introduction of a specific tool, though it does not delve into alternative methods or potential limitations of SAIGE-QTL.

Scientific Rigor, Source Quality, Title Accuracy

The lecture is scientifically rigorous in its presentation of statistical concepts, but it lacks explicit citations to external sources. The title accurately reflects the content. The description provides no links to references, so the only source is the lecture itself. The claim of 48% more eGenes is presented without a reference to the original study, which could be a concern for verification.

153 words

Title / Content Match

The title accurately reflects the content, which focuses on statistical models for single-cell eQTL mapping.

Quality & Reliability

8/10

The lecture is technically sound, presenting established statistical methods (Poisson regression, mixed models) and a specific tool (SAIGE-QTL) with clear explanations. However, it is a single presentation without external validation or references, and some claims (e.g., 48% more eGenes) are not independently verified.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Contribution & Novelties

The lecture provides a clear pedagogical introduction to statistical models for single-cell eQTL mapping, culminating in the SAIGE-QTL framework. Its original contribution is the synthesis of Poisson regression and mixed models into a scalable tool, with demonstrated gains over pseudobulk.

Pour aller plus loin :

  • SAIGE-QTL GitHub — Official repository for the tool.
  • Single-cell eQTL mapping review — A review on single-cell eQTL studies.
  • Poisson regression — Background on Poisson regression.

71 words

Radar Profile

The radar profile shows high scores in technical level and information quality, with moderate scores in quantity and reliability. This indicates a technically dense lecture with solid content but limited breadth and external validation.

Reliability 7/10