Collaborative Research: CISE Crosscutting Small: SCH: Using Explainable AI and Quantum-inspired Computing to Uncover Genetic Insights

About This Grant

The project comprises a strategic collaboration between Florida A&M University (FAMU) and Florida State University (FSU) involving both research and education, aiming to develop capacity in genome research via quantum machine learning (QML) at both universities. Researchers at both universities plan to address current challenges by developing scalable, interpretable, Artificial Intelligence (AI)-based computational tools to interpret single-cell data for understanding the evolution of cancer via single-cell sequencing. Single-cell sequencing - such as single-cell DNA sequencing (scDNA-seq) and single-cell RNA sequencing (scRNA-seq) - has been used to explain and predict how cancer cells evolve. Yet, current phylogenetic tree-inference tools are not scalable to the thousands of cells sequenced by scDNA-seq. In addition, no existing methods can fully automate the process of identifying normal cells in scRNA-seq data, a key step before inferring the cancer evolutionary tree. Consequently, the three main objectives of this project are the following: (1) using quantum-inspired computing to increase the scalability of building a phylogenetic tree on scDNA-seq data; (2) building fully automated, interpretable, deep-learning tools to distinguish between tumor and normal cells on scRNA-seq data; and (3) increasing the number of students studying explainable AI and QML in genome data. Broader-impact aspects of the work include also dissemination of the project via publications as well as open-source code. The project lies at the intersection of AI, genome research, machine learning, and quantum computing. Applying quantum computing to scDNA-seq data is anticipated to advance the frontier of AI-driven analysis using quantum-inspired computers while addressing the critical scalability issue of inferring the phylogenetic tree. The development of interpretable deep-learning methods on scRNA-seq data is expected to reveal the key features that distinguish between tumor and normal cells. It is anticipated that the work will not only contribute to the development of fully automated, interpretable, and robust deep-learning methods for the genomics field but also apply to other domains wherein interpretable data analysis is required, leading to the development of AI-based tools that can be widely used by clinicians and researchers, ultimately contributing to more personalized and effective cancer therapies. In addition, the successful completion of the proposed research will facilitate enhanced collaborative research between FAMU and FSU, train the workforce on AI at FAMU, bring application-driven AI education to FAMU, and accelerate the pace of genomic discovery and cancer-related single-cell data. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.