Finding Better Viral Vectors for Gene Therapy Using Machine Learning

Finding Better Viral Vectors for Gene Therapy Using Machine Learning

Table of Contents


What’s the hold-up on gene therapies? The answer is the challenge of engineering viral vectors for specific, non-toxic delivery of therapeutic genetic information. A new machine learning method called Fit4Function promises to improve the ability of researchers to find AAVs with desirable traits – and the future of gene therapy may not use viruses at all.

Challenges of Gene Therapy Delivery

“Why have so few gene therapies hit the market so far? To date, a mere 22 cell and gene therapy treatments have been approved by the FDA despite the overwhelming promise of gene therapy.” This quote from an article we published in July 2021 still rings true today – in the intervening years, the list of approved therapies has only expanded to 29.

The central challenge of getting gene therapies to market still lies in delivery. Currently, researchers use modified adeno-associated viruses (AAVs), among other types of viruses, to deliver therapeutic genes or gene-editing machinery to cells. The viruses are stripped of their endogenous genetic information which is replaced with a therapeutic payload. 

AAVs were not always an obvious choice for gene therapy delivery. Early pioneers of the field include Ronald Crystal from Weill Cornell Medicine, who had developed a way to replace the A1AT protein in A1AT deficiency. The treatment required weekly infusions of the protein, until 1989, when Crystal collaborated with a former postdoc to develop a “one-time” version of the treatment. By using a virus to deliver the machinery (gene) for a patient’s cells to make the protein themselves, the treatment could last much longer- resulting in one of the first gene therapies. 

Simultaneously in the late 1980s, another pioneer, William Hauswirth at the University of Florida, began investigating gene therapies as a way to treat blindness, eventually developing one of the most successful gene therapy methods to date for blindness caused by LCA2 by delivering a functional copy of the defective sight-related gene to the retina.

While this can be an astoundingly effective approach for simple genetic corrections or local administration (like the retinal-injected gene therapy Luxturna developed by Hauswirth and others), AAVs can’t reach certain tissues and organs very efficiently. To counteract the lack of efficiency, larger doses can be given, but AAVs are often immunogenic and thus can induce a severe immune reaction.

New ML Model Finds “Fit for Function” AAVs

The “rate-limiting step” of delivery research is AAV engineering. Researchers can design and search for viral capsids with desirable traits, like the ability to target a specific tissue type with higher efficiency. Traditional methods are laborious, slow, and have a high error rate, involving random screening of large libraries of AAV capsids in mice. Furthermore, the resulting capsids may demonstrate efficacy in mice but fail to reach similar thresholds in humans.

Ben Deverman, a researcher at the Broad Institute of MIT and Harvard, has been developing ways to improve the search for effective AAVs for therapeutic use for years. Prior to joining the Broad in 2018, Deverman developed a technology at CalTech that could screen large numbers of inactivated AAVs for their therapeutic potential and effectively identify AAVs that could cross the blood-brain barrier in mice and deliver their genetic information to the brain. However, he wanted to build an approach that would find AAVs suitable for human use and turned to machine learning.

In collaboration with a postdoc machine learning specialist, Fatma Elzahraa Eid, the Deverman lab created a machine learning method that can identify AAVs with multiple desirable traits. These include abilities to target certain cell or tissue types while avoiding others, as well as species specificity – AKA the ability to determine which AAVs are most likely to be effective in humans.
Deverman, Eid, and their colleagues describe their new method, called Fit4Function, in a recent preprint highlighted in a news release from the Broad. The method could be a major step forward for accelerating gene therapy delivery research, finding more effective viral delivery vectors with fewer side effects, and highlighting the power lying at the intersection of computer science and biology.

Will the Future of Gene Therapy Still Rely on Viruses?

Technology like Fit4Function is a huge boost for current research on viral vector-based gene therapies. However, there are some limitations inherent to the use of AAVs that no amount of optimization can overcome. Nature Biotechnology recently published a news piece about virus-free gene therapy development, highlighting several companies and their preclinical-stage research.

The fundamental challenges of tissue specificity, safety, and lack of immunogenicity still apply to virus-free therapies, but the primary advantages that could be delivered by non-viral vectors like lipid nanoparticles are lower cost (current viral-based gene therapies are among some of the most expensive treatments in the world) and larger genetic payloads (AAVs have a limit of 4.7 kilobases of material). The field of gene therapy is rich with creativity and innovation and will continue to be a source of interesting research developments for biologists.

Outsourcing Bioinformatics Analysis: How Bridge Informatics Can Help

Shifts in research fields like these are made possible by technological advances making biological data generation, storage, and analysis faster and more accessible than ever before. From pipeline development and software engineering to deploying existing bioinformatics tools, Bridge Informatics can help you on every step of your research journey.

As experts across data types from leading sequencing platforms, we can help you tackle the challenging computational tasks of storing, analyzing, and interpreting genomic and transcriptomic data. Bridge Informatics’ bioinformaticians are trained bench biologists, so they understand the biological questions driving your computational analysis. Click here to schedule a free introductory call with a member of our team.

Jane Cook, Biochemist & Content Writer, Bridge Informatics

Jane Cook, the leading Content Writer for Bridge Informatics, has written over 100 articles on the latest topics and trends for the bioinformatics community. Jane’s broad and deep interdisciplinary molecular biology experience spans developing biochemistry assays to genomics. Prior to joining Bridge, Jane held research assistant roles in biochemistry research labs across a variety of therapeutic areas. While obtaining her B.A. in Biochemistry from Trinity College in Dublin, Ireland, Jane also studied journalism at New York University’s Arthur L. Carter Journalism Institute. As a native Texan, she embraces any challenge that comes her way. Jane hails from Dallas but returns to Ireland any and every chance she gets. If you’re interested in reaching out, please email [email protected] or [email protected].

Share this article with a friend

Create an account to access this functionality.
Discover the advantages