AI and Therapeutic siRNA Design

Despite years of progress, therapeutic siRNA design remains a complex challenge. Our research, spanning nearly two decades, highlights the central importance of thermodynamics in RISC assembly.

Fully modified siRNAs are essential for therapeutic use, but these modifications significantly alter thermodynamic properties. As a result, algorithms trained on unmodified siRNA data often fail to predict the activity of fully modified molecules. Beyond RISC dynamics, efficacy is shaped by features of the target site and mRNA architecture.

Over time, we have built a large dataset capturing the activity of fully chemically modified siRNAs in their native biological context. This dataset offers a unique opportunity to collaborate with leading machine learning experts, such as the Dmitry Korkin Lab at Worcester Polytechnic Institute, to apply artificial intelligence and advanced machine learning approaches to develop models that that predict potent, long-lasting siRNA candidates for therapeutic development.

We are preparing to launch a beta version of a web-based portal offering best-in-class informatics tools, enabling researchers to efficiently identify high-quality siRNA leads and minimize the need for labor-intensive screening and in vivo validation.