EUVIMEDEUVIMED entdecken
ÜbersichtsarbeitOpen Access · CC BY-NC 4.0

Comprehensive review and assessment of multi-species splicing variant prediction: task-specific deep learning models and genomic foundation models

Yinuo Sun, Xiaoyu Wang iD, Yuheng Jia, Seiya Imoto iD, Fuyi Li, Chen Li iD, Jiangning Song iD

Briefings in Bioinformatics · 04.05.2026 · 27 (3)

DOI: 10.1093/bib/bbag329 ↗

Kurz gefasst

Übersichtsarbeit in der Zeitschrift Briefings in Bioinformatics, online erschienen am 21.06.2026. Verfasst von Yinuo Sun und 6 weiteren Autor:innen. Beteiligte Einrichtung: Department of Biochemistry and Molecular Biology, Cancer and Infection Programs, Biomedicine Discovery Institute, Monash University, 23 Innovation Wa…. Hauptthemen laut MEDLINE: Deep Learning, Alternative Splicing, Genomics, Models, Genetic. Die Arbeit nennt 149 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0). Gefördert von: Major and Seed Interdisciplinary Research Projects awarded by Monash University, Australian Research Council Future Fellowship, NHMRC Investigator Fellowship.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Alternative splicing generates transcriptomic and proteomic diversity essential for eukaryotic complexity, yet genetic variants disrupting the splicing code underlie numerous human diseases. Deep learning (DL) models and genomic foundation models (GFMs) have achieved outstanding accuracy for predicting splicing variant effects in humans. However, their transferability to non-human species remains poorly understood, limiting applications in agricultural genomics, comparative biology, and non-model organism research, where experimentally validated variant datasets are limited or lacking. In this study, we comprehensively reviewed 35 computational approaches in terms of their architectural characteristics for splicing site and variant prediction and analysis. We systematically benchmarked the performance of 10 representative models for splicing variant prediction across human, rat, pig, and chicken, including four task-specific DL models and six GFMs, using our manually assembled benchmark datasets. Our benchmarking results revealed a substantial cross-species performance decrease (~21%–33% in the area under the receiver operating characteristic curve - AUROC) using task-specific models from human to non-human species datasets. We then applied a supervised adaptation to frozen GFM embeddings (DNABERT-2, Evo 2, Genos) by adding a lightweight classifier (i.e. a multi-layer perceptron) and reduced the cross-species performance decrease for rat and pig (8.56%–23.84% in AUROC), while performance on chicken was very close to human (decline within 1%, even exceeding by 0.52% when using the Evo 2 embedding). We proposed several directions to improve the prediction performance of splicing variants, including feature representation transfer and multi-modal fusion integrating global context, universal embeddings, and species-aware conditioning. We hope our comprehensive review and performance benchmarking can provide useful computational insights for further advancement of splicing variant prediction.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningAlternative SplicingGenomicsModels, GeneticHumansAnimalsRatsSwineSpecies SpecificityPredictive Learning ModelsComputational BiologyGenetic Variation

Stichwörter der Autor:innen

alternative splicingcross-species predictiondeep learninggenomic foundation modelsperformance benchmarkingsplicing variant prediction

Autor:innen und Institutionen

  • Department of Biochemistry and Molecular Biology, Cancer and Infection Programs, Biomedicine Discovery Institute, Monash University, 23 Innovation Walk, Monash University, Clayton, Victoria, 3800, Australia.
  • School of Computer Science and Engineering, Southeast University, No. 2 SEU Road, Nanjing, Jiangsu Province, 211189, China.
  • Division of Health Medical Intelligence, Human Genome Center, Institute of Medical Science, University of Tokyo, 4-6-1 Shirokane-dai Minato-ku Tokyo 108-8639, Japan.
  • South Australian immunoGENomics Cancer Institute (SAiGENCI), Adelaide University, 4 North Terrace (Corner of George Street and North Terrace) Adelaide, SA 5005, Australia.
  • Department of Medicine, School of Clinical Sciences at Monash Health, Monash University, 246 Clayton Road Clayton, Victoria 3168, Australia.
  • Department of Veterinary Biosciences, Melbourne Veterinary School, The University of Melbourne, Corner of Park Drive and Flemington Road Parkville, Victoria 3052, Australia.

ORCID-Profile: Xiaoyu Wang, Seiya Imoto, Chen Li, Jiangning Song

Förderung

  • Major and Seed Interdisciplinary Research Projects awarded by Monash University
  • Australian Research Council Future Fellowship
  • NHMRC Investigator Fellowship
  • International Joint Usage/Research Center, the Institute of Medical Science, The University of Tokyo
  • National Health and Medical Research Council (NHMRC) of Australia Ideas Grant
  • Australian Research Council

Zitieren

APA

Sun, Y., Wang, X., Jia, Y., Imoto, S., Li, F., Li, C., & Song, J. (2026). Comprehensive review and assessment of multi-species splicing variant prediction: task-specific deep learning models and genomic foundation models. Briefings in Bioinformatics, 27(3). https://doi.org/10.1093/bib/bbag329

Vancouver

Sun Y, Wang X, Jia Y, Imoto S, Li F, Li C, et al. Comprehensive review and assessment of multi-species splicing variant prediction: task-specific deep learning models and genomic foundation models. Briefings in Bioinformatics. 2026;27(3). doi:10.1093/bib/bbag329

BibTeX
@article{Sun2026,
  author = {Sun, Yinuo and Wang, Xiaoyu and Jia, Yuheng and Imoto, Seiya and Li, Fuyi and Li, Chen and Song, Jiangning},
  title = {Comprehensive review and assessment of multi-species splicing variant prediction: task-specific deep learning models and genomic foundation models},
  journal = {Briefings in Bioinformatics},
  year = {2026},
  volume = {27},
  number = {3},
  doi = {10.1093/bib/bbag329}
}
Kostenlos starten →