Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
September 19, 2025Nucleic Acids ResearchOpen Access

Investigating the performance of foundation models on human 3′UTR sequences

View Full Paper
Ask AI
Bookmark
Share

Authors

SVSergey VilovMHMatthias Heinig

Discussion

Loading...

Member takes

Overview

Study uncovers superior results from 3′UTR-specific models in RNA tasks, suggesting the need for genome partitioning.

Key Points

  • Models trained on 3′UTR sequences outperform genome-wide models in three out of four RNA tasks.
  • Evaluation of 3′UTR-specific foundation models reveals their efficacy in RNA-related downstream tasks.
  • Training on a dataset of 3 783 714 3′UTR sequences significantly enhances model performance in functional variant detection.
  • The results highlight the importance of considering genome partitioning in model training and evaluation.

Cite This Study

Vilov et al. (2025) studied this question.

synapsesocial.com/papers/68d464ea31b076d99fa63f5dhttps://doi.org/10.1093/nar/gkaf871
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Investigating the performance of foundation models on human 3’UTR sequences2024 · 5 citations
  2. 2Benchmarking DNA Foundation Models for Genomic Sequence Classification2024 · 14 citations
  3. 3SegmentNT: annotating the genome at single-nucleotide resolution with DNA foundation models2024 · 11 citations
  4. 4Benchmarking DNA foundation models for genomic and genetic tasks2025
  5. 5RNALens: Study on 5' UTR Modeling and Cell-Specificity2025