Not every gene is special: Modelling scale controls the false discovery rate when analysing high-throughput sequencing data
4 affiliations
- 1Department of Biochemistry, Western University, London, Ontario, Canada
- 2Program in Bioinformatics and Genomics, Pennsylvania State University, University Park, Pennsylvania, United States of America
- 3College of Information Science and Technology, Pennsylvania State University, State College, Pennsylvania, United States of America
- 4Department of Statistics, Pennsylvania State University, State College, Pennsylvania, United States of America
Abstract
Differential expression/abundance analyses are commonplace in studies employing high-throughput sequencing (HTS); however different tools often fail to return comparable results when applied to the same dataset. Most tools employ normalisations to attempt to correct for technical variation in the count data. Previously, we demonstrated that these normalisations are often inappropriate due to incorrect assumptions regarding the overall scale (i.e., size) of the biological system in question. In this study, we used a combination of binomial thinning and permutation of sample groupings to produce 100 analysis iterations of 11 RNA-seq and other HTS datasets in which ~5% of all features are expected to be significantly different between groups. This enabled calculation of the false discovery rate (FDR) and sensitivity across the iterations. Our simulations showed that scale misspecification results in poor control of the FDR by several commonly used tools and that, counterintuitively, FDRs increased as the modelled difference between groups increased. Implementing a scale model in ALDEx2 or ALDEx3 ameliorated unacceptably high FDRs; however, there was an inherent trade-off between satisfactory FDR control and high sensitivity- no tool offered both. We established that increasing scale uncertainty also increased the minimum difference between groups required for a feature to be reported as differentially expressed. This phenomenon was consistently observed in disparate types of HTS data and was remarkably consistent. Critically, we leveraged a ‘real-world’, non-permuted analysis of an RNA-seq dataset to demonstrate that the latter effect is not a result of our thinning/permutation approach. Overall, our work highlights the potentially unwitting choice between sensitivity and FDR control that all researchers are making when analysing sequencing data and provides guidance on choosing an appropriate amount of scale uncertainty for the analysis of HTS data. Author summary: Despite attempts to define best practices in differential abundance or expression analysis of high-throughput sequencing data, there is no standardised approach, and a plethora of tools are available to use. Each tool applies different assumptions and methods, which can lead to drastically different results when multiple tools are applied to the same dataset. Our work reproduces multiple recent studies showing that several popular tools exaggerate differences in gene expression or abundance between groups. We build on prior work by showing that these excessively high false discovery rates can be mitigated by using a new approach that models some degree of uncertainty around the total size (or scale) of the biological system under study. We find that scale models offer excellent false-discovery rate control, though at a reduced sensitivity. No tool offers both high sensitivity and a low rate of false positives, highlighting a trade-off that researchers must be cognisant of when analysing HTS data. Accordingly, we extend these observations with guidance on selecting an appropriate degree of scale uncertainty so that analysts can make an informed choice about balancing the ability to detect differences in their data with the confidence that any given difference is real.
Related
- Processing, analysing and modelling kinetic data in the era of high-throughput single-molecule biophysics
- A droplet-based long-read single-cell epigenome sequencing method for the high-throughput profiling of multidimensional regulatory elements within individual cells
- Leveraging high-throughput proteomics and AI-based protein folding to accelerate VAV1 molecular glue discovery.
- High-throughput single-spore genome sequencing of filamentous ascomycete fungi using the SAG-gel platform
- A collaborative study evaluating the detection of virus-infected cells by transcriptomic high-throughput sequencing
BibTeX
@article{santos2026every,
title = {{Not every gene is special: Modelling scale controls the false discovery rate when analysing high-throughput sequencing data}},
author = {Scott J. Dos Santos and Andreea C. Murariu and Justin D. Silverman and Gregory B. Gloor},
journal = {PLOS Computational Biology},
year = {2026},
month = sep,
doi = {10.1371/journal.pcbi.1014728},
url = {https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1014728},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Not every gene is special: Modelling scale controls the false discovery rate when analysing high-throughput sequencing data
AU - Scott J. Dos Santos
AU - Andreea C. Murariu
AU - Justin D. Silverman
AU - Gregory B. Gloor
PY - 2026
DA - 2026/09/08
JO - PLOS Computational Biology
DO - 10.1371/journal.pcbi.1014728
UR - https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1014728
AB - Differential expression/abundance analyses are commonplace in studies employing high-throughput sequencing (HTS); however different tools often fail to return comparable results when applied to the same dataset. Most tools employ normalisations to attempt to correct for technical variation in the count data. Previously, we demonstrated that these normalisations are often inappropriate due to incorrect assumptions regarding the overall scale (i.e., size) of the biological system in question. In this study, we used a combination of binomial thinning and permutation of sample groupings to produce 100 analysis iterations of 11 RNA-seq and other HTS datasets in which ~5% of all features are expected to be significantly different between groups. This enabled calculation of the false discovery rate (FDR) and sensitivity across the iterations. Our simulations showed that scale misspecification results in poor control of the FDR by several commonly used tools and that, counterintuitively, FDRs increased as the modelled difference between groups increased. Implementing a scale model in ALDEx2 or ALDEx3 ameliorated unacceptably high FDRs; however, there was an inherent trade-off between satisfactory FDR control and high sensitivity- no tool offered both. We established that increasing scale uncertainty also increased the minimum difference between groups required for a feature to be reported as differentially expressed. This phenomenon was consistently observed in disparate types of HTS data and was remarkably consistent. Critically, we leveraged a ‘real-world’, non-permuted analysis of an RNA-seq dataset to demonstrate that the latter effect is not a result of our thinning/permutation approach. Overall, our work highlights the potentially unwitting choice between sensitivity and FDR control that all researchers are making when analysing sequencing data and provides guidance on choosing an appropriate amount of scale uncertainty for the analysis of HTS data. Author summary: Despite attempts to define best practices in differential abundance or expression analysis of high-throughput sequencing data, there is no standardised approach, and a plethora of tools are available to use. Each tool applies different assumptions and methods, which can lead to drastically different results when multiple tools are applied to the same dataset. Our work reproduces multiple recent studies showing that several popular tools exaggerate differences in gene expression or abundance between groups. We build on prior work by showing that these excessively high false discovery rates can be mitigated by using a new approach that models some degree of uncertainty around the total size (or scale) of the biological system under study. We find that scale models offer excellent false-discovery rate control, though at a reduced sensitivity. No tool offers both high sensitivity and a low rate of false positives, highlighting a trade-off that researchers must be cognisant of when analysing HTS data. Accordingly, we extend these observations with guidance on selecting an appropriate degree of scale uncertainty so that analysts can make an informed choice about balancing the ability to detect differences in their data with the confidence that any given difference is real.
SN - 1553-7358
ER -