[Paper Review] Diffusion Models in Bioinformatics: A New Wave of Deep Learning Revolution in Action

Zhiye Guo, Jian Liu+5 more|arXiv (Cornell University)|Feb 13, 2023

Gene expression and cancer classificationBiochemistry, Genetics and Molecular Biology12 citations

TL;DR

This review surveys diffusion models and their applications in bioinformatics, outlining foundations and domains from cryo-EM to single-cell analysis, protein design, and drug design.

ABSTRACT

Denoising diffusion models have emerged as one of the most powerful generative models in recent years. They have achieved remarkable success in many fields, such as computer vision, natural language processing (NLP), and bioinformatics. Although there are a few excellent reviews on diffusion models and their applications in computer vision and NLP, there is a lack of an overview of their applications in bioinformatics. This review aims to provide a rather thorough overview of the applications of diffusion models in bioinformatics to aid their further development in bioinformatics and computational biology. We start with an introduction of the key concepts and theoretical foundations of three cornerstone diffusion modeling frameworks (denoising diffusion probabilistic models, noise-conditioned scoring networks, and stochastic differential equations), followed by a comprehensive description of diffusion models employed in the different domains of bioinformatics, including cryo-EM data enhancement, single-cell data analysis, protein design and generation, drug and small molecule design, and protein-ligand interaction. The review is concluded with a summary of the potential new development and applications of diffusion models in bioinformatics.

Motivation & Objective

• Explain the key concepts and theoretical foundations of diffusion modeling (denoising diffusion probabilistic models, noise-conditioned scoring networks, and stochastic differential equations).
• Survey diffusion model applications across bioinformatics domains to map current use-cases and methodologies.
• Summarize challenges, limitations, and potential future directions to guide further development in computational biology.

Proposed method

• Introduce and explain three cornerstone diffusion modeling frameworks: denoising diffusion probabilistic models, noise-conditioned scoring networks, and stochastic differential equations.
• Provide a domain-by-domain survey of diffusion model applications in bioinformatics (cryo-EM data enhancement, single-cell data analysis, protein design and generation, drug and small molecule design, protein-ligand interaction).
• Synthesize insights and discuss potential new developments and applications in bioinformatics.

Experimental results

Research questions

• RQ1 What are the core diffusion modeling frameworks relevant to bioinformatics and how do they differ conceptually and computationally?
• RQ2 How are diffusion models applied across bioinformatics domains such as cryo-EM, single-cell analysis, and molecular design?
• RQ3 What are the current limitations and challenges of using diffusion models in bioinformatics, and what future directions are suggested?

Key findings

• Diffusion models are presented as powerful generative models with broad applicability in bioinformatics.
• The review provides a thorough mapping of diffusion model applications across multiple bioinformatics domains.
• It discusses theoretical foundations and practical considerations to aid further development in computational biology.
• The paper highlights potential new developments and applications of diffusion models in bioinformatics.