Texas A&M Conducts Single-Cell RNA Sequencing Data Analysis with MATLAB
Project Advances Quantum Computing Applications in Life Sciences
“MATLAB provides a seamless and reliable environment for developing quantum computing algorithms. Its quantum computing package offers significant advantages in terms of usability, stability, and portability, and it significantly enhanced our research.”
Key Outcomes
- MATLAB Support Package for Quantum Computing enables easy testing of algorithms locally for rapid development and methods validation
- MATLAB offers advantages over other quantum computing development software, such as Qiskit®, in terms of usability, stability, and portability
- MATLAB Support Package for Quantum Computing is clear, well-structured, and easy to understand for quick implementation of quantum algorithms
In panel C, green lines highlight the 18 genes identified by both LASSO- and QUBO-based methods, while the unhighlighted lines represent genes identified exclusively by the QUBO-based method.
At Texas A&M’s College of Veterinary Medicine and Biomedical Sciences, Professor James Cai is leading an interdisciplinary project that uses quantum computing to analyze single-cell gene expression data. Professor Cai’s team is using gate-based quantum computing to build networks that show how genes regulate each other. They are also using a method called simulated quantum annealing (QA) with quadratic unconstrained binary optimization (QUBO) to choose important genes from scRNA-seq data that are involved in how cells change and develop.
MATLAB® Tabu search implementation is used for simulated annealing to solve QUBO problems. For example, a QUBO-based feature selection algorithm identified 10 apparently nonlinear gene interactions from the 50 features initially selected out of 5,000. Among these 50 features, only 18 overlapped with those identified using the comparative LASSO selection method. This demonstrates that the QUBO-QA approach captures not only core linear gene expressions but also uncovers intricate nonlinear gene expression patterns.
Statistics and Machine Learning Toolbox™ is used extensively throughout the project, particularly in data processing workflows. Professor Cai developed scGEAToolbox to facilitate the analyses of scRNA-seq data in the MATLAB environment. It contains a comprehensive set of functions for data normalization, feature selection, cell clustering, cell type annotation, pseudotime analysis, gene network construction, virtual gene knockout analysis, and cell-cell communication analysis. Curve Fitting Toolbox™, Parallel Computing Toolbox™, and Image Processing Toolbox™ are also used for visualization and to complement analyses.
The project is a landmark example of how quantum computing can be applied in life sciences. With MATLAB, collaborators can replicate and build upon this initial work to further the application of quantum computing, specifically for transcriptomics research. In the long run, it may profoundly change how quantum computing is used to develop personalized diagnoses and treatments based on individual genetic data.
Related Resources
