Exploring transcriptomic and genomic latent variable correction approaches in differential expression analysis
Abstract
Differential expression analysis is a central tool for studying the biological processes altered in human diseases via transcriptomic signatures. However, transcriptomic datasets are systematically confounded by latent variables from two distinct sources: unmeasured technical and biological heterogeneity within the expression data, and expression differences driven by population stratification. Correction using expression-based surrogate variables (SVs) and genotype-based principal components (PCs) addresses these sources independently, yet no study has directly evaluated their combined use against either method alone within a differential expression framework. In this study we hypothesised that simultaneously including both correction layers would produce more biologically valid and reproducible results than either approach alone, and tested this in two independent post-mortem RNA-seq datasets of amyotrophic lateral sclerosis (ALS) cases and controls with paired genotype data. Four nested differential expression models (corrected for PC-only, SV-only, both SV and PC, and neither PCs nor SVs) were evaluated across the KCLBB (96 cases and 52 controls) and ALS Consortium (272 cases and 35 controls) datasets. Models were evaluated on cross-dataset effect size concordance, cross-dataset replicability quantified by the Jaccard Similarity Index, and biological recall against a curated reference set of 66 known ALS genes. The combined SV + PC framework outperformed simpler models across most metrics. Replicability improved nearly ten-fold compared to the non-corrected model, (Jaccard index: 2.28% to 19.5%), and the combined framework exhibited a statistically significant 2.2% gain over the SV-only model. Biological recall doubled compared to SV correction alone. Effect size magnitude consistency was preserved, though a reduction in Spearman’s rank correlation indicates PC correction refines magnitude rather than gene rankings. These findings remained generally robust to genotype PC number and sequencing platform adjustment, with more variable performance observed under ancestry restriction and upon extension to an independent tissue. This study found that SVs and genotype PCs address non-redundant sources of confounding, and we recommend their combined use as standard practice in differential expression analysis where paired genotype data are available, with the greatest benefit expected in ancestrally diverse cohorts. Notably PCs capturing population structure can also be derived directly from RNA-seq data, potentially extending this framework’s applicability to studies lacking paired genotype data. Although this analysis was restricted to ALS datasets, we expect these findings to generalise to other traits.
Related
- Integrating zero-inflation correction and transcriptional kinetics for single-cell transcriptomic analysis
- Assessing the impact of parental linear gene normalization on the performance of statistical models for circular RNA differential expression analysis
- Differential Expression of Cytotoxic-Associated Markers in Circulating CD4+CD28− T Cells and NK Cells in Rheumatoid Arthritis
- Cell DiffErential Expression by Pooling (CellDEEP) highlights issues in differential gene expression in scRNA-seq
- TSPO and LRBA Expression Is Associated with Acute Renal Allograft Rejection: An Integrative Transcriptomic Analysis
BibTeX
@article{appulingam2026exploring,
title = {{Exploring transcriptomic and genomic latent variable correction approaches in differential expression analysis}},
author = {Yadusayan Appulingam and John Jammal and Aminah Ali and Simon Topp and NYGC ALS Consortium and Hemali Phatnani and Justin Kwan and Dhruv Sareen and James R. Broach and Zachary Simmons and Ximena Arcila-Londono and Edward B. Lee and Vivianna M. Van Deerlin and Neil A. Shneider and Ernest Fraenkel and Lyle W. Ostrow and Frank Baas and Noah Zaitlen and James D. Berry and Andrea Malaspina and Pietro Fratta and Gregory A. Cox and Leslie M. Thompson and Steve Finkbeiner and Efthimios Dardiotis and Timothy M. Miller and Siddharthan Chandran and Suvankar Pal and Eran Hornstein and Daniel J. MacGowan and Terry Heiman-Patterson and Molly G. Hammell and Nikolaos. A. Patsopoulos and Oleg Butovsky and Joshua Dubnau and Avindra Nath and Robert Bowser and Matthew Harms and Eleonora Aronica and Mary Poss and Jennifer Phillips-Cremins and John Crary and Nazem Atassi and Dale J. Lange and Darius J. Adams and Leonidas Stefanis and Marc Gotkine and Robert H. Baloh and Suma Babu and Towfique Raj and Sabrina Paganoni and Ophir Shalem and Colin Smith and Bin Zhang and Justin Kwan and Thomas Blanchard and Brent Harris and Iris Broce and Vivian Drory and John Ravits},
journal = {Scientific Reports},
year = {2026},
month = oct,
doi = {10.1038/s41598-026-69619-8},
url = {https://www.nature.com/articles/s41598-026-69619-8},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Exploring transcriptomic and genomic latent variable correction approaches in differential expression analysis
AU - Yadusayan Appulingam
AU - John Jammal
AU - Aminah Ali
AU - Simon Topp
AU - NYGC ALS Consortium
AU - Hemali Phatnani
AU - Justin Kwan
AU - Dhruv Sareen
AU - James R. Broach
AU - Zachary Simmons
AU - Ximena Arcila-Londono
AU - Edward B. Lee
AU - Vivianna M. Van Deerlin
AU - Neil A. Shneider
AU - Ernest Fraenkel
AU - Lyle W. Ostrow
AU - Frank Baas
AU - Noah Zaitlen
AU - James D. Berry
AU - Andrea Malaspina
AU - Pietro Fratta
AU - Gregory A. Cox
AU - Leslie M. Thompson
AU - Steve Finkbeiner
AU - Efthimios Dardiotis
AU - Timothy M. Miller
AU - Siddharthan Chandran
AU - Suvankar Pal
AU - Eran Hornstein
AU - Daniel J. MacGowan
AU - Terry Heiman-Patterson
AU - Molly G. Hammell
AU - Nikolaos. A. Patsopoulos
AU - Oleg Butovsky
AU - Joshua Dubnau
AU - Avindra Nath
AU - Robert Bowser
AU - Matthew Harms
AU - Eleonora Aronica
AU - Mary Poss
AU - Jennifer Phillips-Cremins
AU - John Crary
AU - Nazem Atassi
AU - Dale J. Lange
AU - Darius J. Adams
AU - Leonidas Stefanis
AU - Marc Gotkine
AU - Robert H. Baloh
AU - Suma Babu
AU - Towfique Raj
AU - Sabrina Paganoni
AU - Ophir Shalem
AU - Colin Smith
AU - Bin Zhang
AU - Justin Kwan
AU - Thomas Blanchard
AU - Brent Harris
AU - Iris Broce
AU - Vivian Drory
AU - John Ravits
PY - 2026
DA - 2026/10/11
JO - Scientific Reports
DO - 10.1038/s41598-026-69619-8
UR - https://www.nature.com/articles/s41598-026-69619-8
AB - Differential expression analysis is a central tool for studying the biological processes altered in human diseases via transcriptomic signatures. However, transcriptomic datasets are systematically confounded by latent variables from two distinct sources: unmeasured technical and biological heterogeneity within the expression data, and expression differences driven by population stratification. Correction using expression-based surrogate variables (SVs) and genotype-based principal components (PCs) addresses these sources independently, yet no study has directly evaluated their combined use against either method alone within a differential expression framework. In this study we hypothesised that simultaneously including both correction layers would produce more biologically valid and reproducible results than either approach alone, and tested this in two independent post-mortem RNA-seq datasets of amyotrophic lateral sclerosis (ALS) cases and controls with paired genotype data. Four nested differential expression models (corrected for PC-only, SV-only, both SV and PC, and neither PCs nor SVs) were evaluated across the KCLBB (96 cases and 52 controls) and ALS Consortium (272 cases and 35 controls) datasets. Models were evaluated on cross-dataset effect size concordance, cross-dataset replicability quantified by the Jaccard Similarity Index, and biological recall against a curated reference set of 66 known ALS genes. The combined SV + PC framework outperformed simpler models across most metrics. Replicability improved nearly ten-fold compared to the non-corrected model, (Jaccard index: 2.28% to 19.5%), and the combined framework exhibited a statistically significant 2.2% gain over the SV-only model. Biological recall doubled compared to SV correction alone. Effect size magnitude consistency was preserved, though a reduction in Spearman’s rank correlation indicates PC correction refines magnitude rather than gene rankings. These findings remained generally robust to genotype PC number and sequencing platform adjustment, with more variable performance observed under ancestry restriction and upon extension to an independent tissue. This study found that SVs and genotype PCs address non-redundant sources of confounding, and we recommend their combined use as standard practice in differential expression analysis where paired genotype data are available, with the greatest benefit expected in ancestrally diverse cohorts. Notably PCs capturing population structure can also be derived directly from RNA-seq data, potentially extending this framework’s applicability to studies lacking paired genotype data. Although this analysis was restricted to ALS datasets, we expect these findings to generalise to other traits.
SN - 2045-2322
ER -