Extending Sequence Length is Not All You Need: Effective Integration of Multimodal Signals for Gene Expression Prediction

2026-02-24

AI Summary

Existing methods for gene expression prediction often rely excessively on extending DNA sequence length while neglecting the effective integration of proximal multimodal epigenomic signals, thereby introducing spurious associations. To address this limitation, this work proposes Prism, a novel framework that systematically models the distinct biological roles of diverse epigenomic signals. Prism characterizes background chromatin states through high-dimensional feature combinations to differentiate functional regulatory signals from nonspecific background noise and incorporates backdoor adjustment—a causal inference technique—to mitigate confounding effects. Remarkably, using only short DNA sequences, Prism achieves state-of-the-art performance in gene expression prediction, demonstrating that precise integration of multimodal epigenomic signals is superior to merely increasing sequence length.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Graphical Models
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Gene expression prediction, which predicts mRNA expression levels from DNA sequences, presents significant challenges. Previous works often focus on extending input sequence length to locate distal enhancers, which may influence target genes from hundreds of kilobases away. Our work first reveals that for current models, long sequence modeling can decrease performance. Even carefully designed algorithms only mitigate the performance degradation caused by long sequences. Instead, we find that proximal multimodal epigenomic signals near target genes prove more essential. Hence we focus on how to better integrate these signals, which has been overlooked. We find that different signal types serve distinct biological roles, with some directly marking active regulatory elements while others reflect background chromatin patterns that may introduce confounding effects. Simple concatenation may lead models to develop spurious associations with these background patterns. To address this challenge, we propose Prism, a framework that learns multiple combinations of high-dimensional epigenomic features to represent distinct background chromatin states and uses backdoor adjustment to mitigate confounding effects. Our experimental results demonstrate that proper modeling of multimodal epigenomic signals achieves state-of-the-art performance using only short sequences for gene expression prediction.

Problem

• gene expression prediction
• multimodal epigenomic signals
• confounding effects
• background chromatin patterns
• spurious associations

Innovation

• multimodal integration
• gene expression prediction
• epigenomic signals
• confounding mitigation
• backdoor adjustment