CoChat
Sign Up
Cited by 128Genomics and Chromatin DynamicsGene Regulatory Network AnalysisGenetic Associations and Epidemiology

Current sequence-based models capture gene expression determinants in promoters but mostly ignore distal enhancers

Open access
Karollus, Alexander, Mauermeier, Thomas, Gagneur, Julien·2023·Genome Biology
DOI 10.1186/s13059-023-02899-9PMC PMC10045630
Open this paper in CoChat — free accountOpen PDF

Abstract

Background: The largest sequence-based models of transcription control to date are obtained by predicting genome-wide gene regulatory assays across the human genome. This setting is fundamentally correlative, as those models are exposed during training solely to the sequence variation between human genes that arose through evolution, questioning the extent to which those models capture genuine causal signals. Results: Here we confront predictions of state-of-the-art models of transcription regulation against data from two large-scale observational studies and five deep perturbation assays. The most advanced of these sequence-based models, Enformer, by and large, captures causal determinants of human promoters. However, models fail to capture the causal effects of enhancers on expression, notably in medium to long distances and particularly for highly expressed promoters. More generally, the predicted impact of distal elements on gene expression predictions is small and the ability to correctly integrate long-range information is significantly more limited than the receptive fields of the models suggest. This is likely caused by the escalating class imbalance between actual and candidate regulatory elements as distance increases. Conclusions: Our results suggest that sequence-based models have advanced to the point that in silico study of promoter regions and promoter variants can provide meaningful insights and we provide practical guidance on how to use them. Moreover, we foresee that it will require significantly more and particularly new kinds of data to train models accurately accounting for distal elements.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1186/s13059-023-02899-9

Cites

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Performance decays with distance to the target gene across all eQTL tasks.

    “As previously reported26, performance decays with distance to the target gene across all eQTL tasks (Extended Data Fig. 5f).”

    From Advancing regulatory variant effect prediction with AlphaGenome (2026) · cites this paper as [26]

    In Results, ‘Most distal eQTLs do not have a meaningful impact on expression predictions’ (Fig. 6A–B), the paper reports that Enformer’s predicted eQTL effects decay rapidly with distance to the target gene’s TSS. It does not establish that performance decays across all eQTL tasks; it also reports no systematic distance-related decay in the measured eQTL effect sizes. (LLM, conf=0.86)

    Experimental — accuracy not yet measured

  • There is a general decreasing trend in average predicted percent expression change with TSS distance for both positive and negative examples.

    “In line with recent work66, we find a general decreasing trend in average predicted percent expression change with TSS distance for both positive and negative examples (Supplementary Fig. 6a).”

    From Predicting RNA-seq coverage from DNA sequence as a unifying model of gene regulation (2025) · cites this paper as [66]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

No citation lists for this paper

No citation lists for this paper yet — checked again daily.

References 47 · Cited by 128

Research works printed but not linked: 46

  1. [2]Eraslan B Wang D Gusic M Prokisch H Hallström BM Uhlén M Quantification and discovery of sequence determinants of protein-per-mRNA amount in 29 human tissues Mol Syst Biol 2019 15 e8513 10.15252/msb.20188513 30777893 PMC6379048 · DOI
  2. [3]Agarwal V Shendure J Predicting mRNA abundance directly from genomic sequence using deep convolutional neural networks Cell Rep 2020 31 107663 10.1016/j.celrep.2020.107663 32433972 · DOI · Paper page
  3. [4]Kelley DR Reshef YA Bileschi M Belanger D McLean CY Snoek J Sequential regulatory activity prediction across chromosomes with convolutional neural networks Genome Res 2018 28 739 50 10.1101/gr.227819.117 29588361 PMC5932613 · DOI · Paper page
  4. [5]Kelley DR Cross-species regulatory sequence activity prediction PLoS Comput Biol. 2020 16 e1008050 10.1371/journal.pcbi.1008050 32687525 PMC7392335 · DOI · Paper page
  5. [6]ENCODE Project Consortium An integrated encyclopedia of DNA elements in the human genome Nature 2012 489 57 74 10.1038/nature11247 22955616 PMC3439153 · DOI · Paper page
  6. [7]Luo Y Hitz BC Gabdank I Hilton JA Kagda MS Lam B New developments on the Encyclopedia of DNA Elements (ENCODE) data portal Nucleic Acids Res 2020 48 D882 9 10.1093/nar/gkz1062 31713622 PMC7061942 · DOI
  7. [8]Noguchi S Arakawa T Fukuda S Furuno M Hasegawa A Hori F FANTOM5 CAGE profiles of human and mouse samples Sci Data 2017 4 170112 10.1038/sdata.2017.112 28850106 PMC5574368 · DOI · Paper page
  8. [9]Zhou J Theesfeld CL Yao K Chen KM Wong AK Troyanskaya OG Deep learning sequence-based ab initio prediction of variant effects on expression and disease risk Nat Genet 2018 50 1171 9 10.1038/s41588-018-0160-6 30013180 PMC6094955 · DOI · Paper page
  9. [10]Zhou J Troyanskaya OG Predicting effects of noncoding variants with deep learning-based sequence model Nat Methods 2015 12 931 4 10.1038/nmeth.3547 26301843 PMC4768299 · DOI · Paper page
  10. [11]Avsec Ž Agarwal V Visentin D Ledsam JR Grabska-Barwinska A Taylor KR Effective gene expression prediction from sequence by integrating long-range interactions Nat Methods 2021 18 1196 203 10.1038/s41592-021-01252-x 34608324 PMC8490152 · DOI · Paper page
  11. [12]Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, et al. Attention is all you need. 2017. Cited 2022 Sep 13; Available from: 10.48550/arXiv.1706.03762. · DOI · Paper page
  12. [13]Bergman DT Jones TR Liu V Ray J Jagoda E Siraj L Compatibility rules of human enhancer and promoter sequences Nature 2022 607 176 84 10.1038/s41586-022-04877-w 35594906 PMC9262863 · DOI
  13. [14]Kircher M Xiong C Martin B Schubach M Inoue F Bell RJA Saturation mutagenesis of twenty disease-associated regulatory elements at single base-pair resolution Nat Commun 2019 10 1 15 10.1038/s41467-019-11526-w 31395865 PMC6687891 · DOI · Paper page
  14. [15]Lonsdale J Thomas J Salvatore M Phillips R Lo E Shad S The Genotype-Tissue Expression (GTEx) project Nat Genet 2013 45 580 5 10.1038/ng.2653 23715323 PMC4010069 · DOI · Paper page
  15. [16]Cardoso-Moreira M Halbert J Valloton D Velten B Chen C Shao Y Gene expression across mammalian organ development Nature 2019 571 505 9 10.1038/s41586-019-1338-5 31243369 PMC6658352 · DOI · Paper page
  16. [17]GTEx Consortium The GTEx Consortium atlas of genetic regulatory effects across human tissues Science 2020 369 1318 30 10.1126/science.aaz1776 32913098 PMC7737656 · DOI · Paper page
  17. [18]Cunningham F Allen JE Allen J Alvarez-Jarreta J Amode MR Armean IM Ensembl 2022 Nucleic Acids Res 2022 50 D988 95 10.1093/nar/gkab1049 34791404 PMC8728283 · DOI · Paper page
  18. [19]Xia B Yan Y Baron M Wagner F Barkley D Chiodin M Widespread transcriptional scanning in the testis modulates gene evolution rates Cell 2020 180 248 62.e21 10.1016/j.cell.2019.12.015 31978344 PMC7891839 · DOI
  19. [20]Gasperini M Hill AJ McFaline-Figueroa JL Martin B Kim S Zhang MD A genome-wide framework for mapping gene regulation via cellular genetic screens Cell 2019 176 1516 10.1016/j.cell.2019.02.027 30849375 · DOI
  20. [21]Fulco CP Nasser J Jones TR Munson G Bergman DT Subramanian V Activity-by-contact model of enhancer-promoter regulation from thousands of CRISPR perturbations Nat Genet 2019 51 1664 9 10.1038/s41588-019-0538-0 31784727 PMC6886585 · DOI · Paper page
  21. [22]Võsa U Claringbould A Westra H-J Bonder MJ Deelen P Zeng B Large-scale cis- and trans-eQTL analyses identify thousands of genetic loci and polygenic scores that regulate blood gene expression Nat Genet 2021 53 1300 10 10.1038/s41588-021-00913-z 34475573 PMC8432599 · DOI · Paper page
  22. [23]Ferraro NM, Strober BJ, Einson J, Abell NS, Aguet F, Barbeira AN, et al. Transcriptomic signatures across human tissues identify functional rare genetic variation. Science. 2020;369. Available from: 10.1126/science.aaz5900. PMC7646251 32913073 · DOI
  23. [24]ENCODE Project Consortium Moore JE Purcaro MJ Pratt HE Epstein CB Shoresh N Expanded encyclopaedias of DNA elements in the human and mouse genomes Nature 2020 583 699 710 10.1038/s41586-020-2493-4 32728249 PMC7410828 · DOI · Paper page
  24. [27]Kerimov N Hayhurst JD Peikova K Manning JR Walter P Kolberg L A compendium of uniformly processed human gene expression and splicing quantitative trait loci Nat Genet 2021 53 1290 9 10.1038/s41588-021-00924-w 34493866 PMC8423625 · DOI · Paper page
  25. [28]Hong CKY Cohen BA Genomic environments scale the activities of diverse core promoters Genome Res 2022 32 85 96 10.1101/gr.276025.121 34961747 PMC8744677 · DOI
  26. [29]Zuin J, Roth G, Zhan Y, Cramard J, Redolfi J, Piskadlo E, et al. Nonlinear control of transcription through enhancer–promoter interactions. Nature. 2022. 571–7. Available from: 10.1038/s41586-022-04570-y PMC9021019 35418676 · DOI · Paper page
  27. [30]Abramov S Boytsov A Bykova D Penzar DD Yevshin I Kolmykov SK Landscape of allele-specific transcription factor binding in the human genome Nat Commun 2021 12 2751 10.1038/s41467-021-23007-0 33980847 PMC8115691 · DOI
  28. [31]Chen KM Wong AK Troyanskaya OG Zhou J A sequence-based global map of regulatory activity for deciphering human genetics Nat Genet 2022 54 940 9 10.1038/s41588-022-01102-2 35817977 PMC9279145 · DOI · Paper page
  29. [35]Bogard N Linder J Rosenberg AB Seelig G A deep neural network for predicting and engineering alternative polyadenylation Cell 2019 178 91 106.e23 10.1016/j.cell.2019.04.046 31178116 PMC6599575 · DOI · Paper page
  30. [36]Vaishnav ED de Boer CG Molinet J Yassour M Fan L Adiconis X The evolution, evolvability and engineering of gene regulatory DNA Nature 2022 603 455 63 10.1038/s41586-022-04506-6 35264797 PMC8934302 · DOI
  31. [38]Fudenberg G Kelley DR Pollard KS Predicting 3D genome folding from DNA sequence with Akita Nat Methods 2020 17 1111 7 10.1038/s41592-020-0958-x 33046897 PMC8211359 · DOI
  32. [39]Schwessinger R Gosden M Downes D Brown RC Oudelaar AM Telenius J DeepC: predicting 3D genome folding using megabase-scale transfer learning Nat Methods 2020 17 1118 24 10.1038/s41592-020-0960-3 33046896 PMC7610627 · DOI
  33. [40]Hecker D, Behjati Ardakani F, Karollus A, Gagneur J, Schulz MH. The adapted Activity-By-Contact model for enhancer-gene assignment and its application to single-cell data. Bioinformatics [Internet]. 2023; Available from: 10.1093/bioinformatics/btad062 PMC9931646 36708003 · DOI
  34. [41]Avsec Ž Kreuzhuber R Israeli J Xu N Cheng J Shrikumar A The Kipoi repository accelerates community exchange and reuse of predictive models for genomics Nat Biotechnol 2019 37 592 600 10.1038/s41587-019-0140-0 31138913 PMC6777348 · DOI
  35. [42]Athar A Füllgrabe A George N Iqbal H Huerta L Ali A ArrayExpress update - from bulk to single-cell expression data Nucleic Acids Res 2019 47 D711 5 10.1093/nar/gky964 30357387 PMC6323929 · DOI
  36. [44]Stovner EB Sætrom P PyRanges: efficient comparison of genomic intervals in Python Bioinformatics 2020 36 918 9 10.1093/bioinformatics/btz615 31373614 · DOI
  37. [45]Seabold S, Perktold J. Statsmodels: econometric and statistical modeling with python. Proceedings of the 9th Python in Science Conference. SciPy; 2010. Available from: https://conference.scipy.org/proceedings/scipy2010/seabold.html
  38. [46]Karollus A, Mauermeier T, Gagneur J. Current sequence-based models capture gene expression determinants in promoters but mostly ignore distal enhancers. Datasets. Zenodo. https://zenodo.org/record/7076228 (2022). PMC10045630 36973806
  39. [48]Weingarten-Gabbay S, Nir R, Lubliner S, Sharon E et al. Systematic interrogation of human promoters. Datasets. Gene Expression Omnibus, GSE118242. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE118242 (2018) PMC6360817 30622120
  40. [49]Bergman DT, Jones TR, Liu V, Ray J et al. Compatibility rules of human enhancer and promoter sequences. Datasets. Gene Expression Omnibus, GSE184426. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE184426 (2021)
  41. [50]Kircher M, Xiong C, Martin B, Schubach M et al. Saturation mutagenesis of twenty disease-associated regulatory elements at single base-pair resolution. Datasets. Gene Expression Omnibus, GSE126550. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE126550 (2019) PMC6687891 31395865
  42. [51]Fulco CP, Nasser J, Jones TR, Munson G et al. Activity-by-contact model of enhancer-promoter regulation from thousands of CRISPR perturbations. Datasets. Gene Expression Omnibus, GSE118912. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE118912 (2019) PMC6886585 31784727
  43. [52]Gasperini M, Hill AJ, McFaline-Figueroa JL, Martin B et al. A genome-wide framework for mapping gene regulation via cellular genetic screens. Datasets. Gene Expression Omnibus, GSE120861. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE120861 (2019) PMC6690346 30612741
  44. [53]Kerimov N, Hayhurst JD, Peikova K, Manning JR, Walter P, Kolberg L, et al. A compendium of uniformly processed human gene expression and splicing quantitative trait loci. Datasets. EMBL-EBI. http://ftp.ebi.ac.uk/pub/databases/spot/eQTL/credible_sets/ (2021) PMC8423625 34493866
  45. [54]Hong CKY, Cohen BA. Genomic environments scale the activities of diverse core promoters. Datasets. Gene Expression Omnibus, GSE173678. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE173678 (2021) PMC8744677 34961747
  46. [56]Karollus A, Mauermeier T, Gagneur J. Code for the paper “Current sequence-based models capture gene expression determinants in promoters but mostly ignore distal enhancers”. Zenodo. https://zenodo.org/record/7613255 (2023). PMC10045630 36973806

Authors
Karollus, A., Mauermeier, T., & Gagneur, J.
Affiliations
  • Technical University of Munich (DE)
  • Helmholtz Munich (DE)
Corresponding author
  • Karollus, Alexander
  • Gagneur, Julien
ORCID
Venue
Genome Biology
Publisher
Springer Science and Business Media LLC
Published
2023-03-27
Volume
24
Issue
1
Pages
56-56
ISSN
  • 1465-6906 (ISSN-L)
  • 1474-760X
DOI
10.1186/s13059-023-02899-9
PMC
PMC10045630
OpenAlex
W4361001523
Cited by
128
References
47
Citations by year
2026: 37 · 2025: 27 · 2024: 37 · 2023: 26 · 2022: 1
FWCI
11.31
Type
journal article
Language
en
Open access
Licence
Locations
  • Genome biology (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • mediaTUM (Technical University of Munich) (submittedVersion)
  • PuSH - Publication Server of Helmholtz Zentrum München (submittedVersion)
Funders
  • Bundesministerium für Bildung und Forschung (031L0174A)
  • Technische Universität München
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Chromatin Dynamics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Gene Regulatory Network Analysis
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Genetic Associations and Epidemiology
Keywords
transcriptional regulation, distal enhancers, promoter polymorphism, Enformer, causal inference, class imbalance
MeSH terms
Enhancer Elements, Genetic, Gene Expression Regulation, Humans, Promoter Regions, Genetic, Gene Expression, Genomics

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review