GeneEmbedLLM: Harnessing Large Language Models for Gene Functional Embedding and Association Prediction
- 1Guangzhou National Laboratory
Abstract
Effective and high-quality gene function representations are crucial for bioinformatics tasks. While biological knowledge about genes is inherently textual, encompassing functional descriptions and pathway annotations, current methods often rely on single-source, non-textual data, failing to capture the rich semantic information in multi-source gene annotations. In this paper, we propose GeneEmbedLLM, a Large Language Model (LLM)-powered framework for generating universal, semantically-aware gene embeddings. The proposed approach advances the gene representation learning through multi-source structured text integration from five authoritative databases and attribute-aware attention based on Longformer architecture and contrastive learning. Experimental results on diverse downstream tasks, ranging from molecular interactions and phenotypic dependencies to epigenetic regulation, confirm the validity and effectiveness of our proposed approach, where competitive performance is achieved across all evaluated benchmarks. We further validate the quality of the generated embeddings through comprehensive analyses, including embedding visualizations, intra-pathway similarity assessments, ablation study, and parameter analysis.
Related
- Harnessing engineered laccase and functional carrier materials for efficient dye degradation: From mechanism to application.
- Harnessing variation in Drosophila melanogaster reveals different routes to increased infection susceptibility over age and sex-dependent association with lifespan
- ABCP_finder: A Transformer Embedding-Based Prediction of Anti-Breast Cancer Peptides
- Genome-Wide Association and Genomic Prediction of Powdery Mildew and CYSDV Resistance in Melon
- Toward a Transferable Representation of Protein Dynamics for Functional Prediction and Mechanistic Discovery
BibTeX
@misc{zhang2026geneembedllm,
title = {{GeneEmbedLLM: Harnessing Large Language Models for Gene Functional Embedding and Association Prediction}},
author = {z. zhang and l. yang and z. he and j. ding and Y. Li and R. Shen},
howpublished = {bioRxiv},
year = {2026},
month = nov,
doi = {10.64898/2026.11.05.764758},
url = {https://www.biorxiv.org/content/10.64898/2026.11.05.764758v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - GeneEmbedLLM: Harnessing Large Language Models for Gene Functional Embedding and Association Prediction
AU - z. zhang
AU - l. yang
AU - z. he
AU - j. ding
AU - Y. Li
AU - R. Shen
PY - 2026
DA - 2026/11/05
PB - bioRxiv
DO - 10.64898/2026.11.05.764758
UR - https://www.biorxiv.org/content/10.64898/2026.11.05.764758v1
AB - Effective and high-quality gene function representations are crucial for bioinformatics tasks. While biological knowledge about genes is inherently textual, encompassing functional descriptions and pathway annotations, current methods often rely on single-source, non-textual data, failing to capture the rich semantic information in multi-source gene annotations. In this paper, we propose GeneEmbedLLM, a Large Language Model (LLM)-powered framework for generating universal, semantically-aware gene embeddings. The proposed approach advances the gene representation learning through multi-source structured text integration from five authoritative databases and attribute-aware attention based on Longformer architecture and contrastive learning. Experimental results on diverse downstream tasks, ranging from molecular interactions and phenotypic dependencies to epigenetic regulation, confirm the validity and effectiveness of our proposed approach, where competitive performance is achieved across all evaluated benchmarks. We further validate the quality of the generated embeddings through comprehensive analyses, including embedding visualizations, intra-pathway similarity assessments, ablation study, and parameter analysis.
ER -