GENERator: A Long-Context Generative Genomic Foundation Model

2025-02-11

AI Summary

Current large language models (LLMs) face limitations in genomic sequence analysis due to constrained context length, insufficient parameter scale, and lack of biological priors, resulting in poor robustness and generalization. To address this, we introduce the first generative genomic foundation model—1.2B parameters, trained on 386 billion base pairs of eukaryotic DNA—with a 98-kilobase context window. Our method enables end-to-end, central-dogma-aligned protein-coding sequence generation, supporting prompt-driven, function-oriented design. We incorporate biologically grounded pretraining objectives, genome-specific tokenization, and optimized positional encoding. The model achieves state-of-the-art performance across multiple genomic benchmarks. Experimental validation confirms its ability to generate protein-coding sequences with native-like structural properties and to controllably design highly active promoters—demonstrating a 3.2-fold increase in transcriptional activity over baselines.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Natural Language Processing: Generation
• Computer Vision: Large Vision Models

Application Category

• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Advancements in DNA sequencing technologies have significantly improved our ability to decode genomic sequences. However, the prediction and interpretation of these sequences remain challenging due to the intricate nature of genetic material. Large language models (LLMs) have introduced new opportunities for biological sequence analysis. Recent developments in genomic language models have underscored the potential of LLMs in deciphering DNA sequences. Nonetheless, existing models often face limitations in robustness and application scope, primarily due to constraints in model structure and training data scale. To address these limitations, we present GENERator, a generative genomic foundation model featuring a context length of 98k base pairs (bp) and 1.2B parameters. Trained on an expansive dataset comprising 386B bp of eukaryotic DNA, the GENERator demonstrates state-of-the-art performance across both established and newly proposed benchmarks. The model adheres to the central dogma of molecular biology, accurately generating protein-coding sequences that translate into proteins structurally analogous to known families. It also shows significant promise in sequence optimization, particularly through the prompt-responsive generation of promoter sequences with specific activity profiles. These capabilities position the GENERator as a pivotal tool for genomic research and biotechnological advancement, enhancing our ability to interpret and predict complex biological systems and enabling precise genomic interventions.

Problem

• Predicting and interpreting complex genomic sequences
• Overcoming limitations in robustness and application scope of genomic models
• Enhancing sequence optimization and precise genomic interventions

Innovation

• Long-context generative model
• 1.2B parameters
• 386B bp training dataset