Original title: Adaptive Regularized Tri-Factor Non-Negative Matrix Factorization for Cell Type Deconvolution
Authors: Tianyi Liu,Quefeng Li,Xiaojing Zheng,Fei Zou
In this article, the authors discuss the importance of accurately identifying different cell types from bulk gene expression data for understanding disease and physiological states. They point out that current methods either require complete gene expression signatures or ignore important biological information, leading to biased estimates of cell proportions. Additionally, these methods do not utilize reference information from external studies. To address these challenges, the authors propose a new method called ARTdeConv. They rigorously prove the numerical convergence of their algorithm and demonstrate through simulations that ARTdeConv outperforms existing reference-free methods. They also validate their method in a real dataset from a flu vaccine study, showing a high correlation with flow cytometry measurements. The authors have made their method available as an R package on GitHub and declare no competing interests.
Original article: https://www.biorxiv.org/content/10.1101/2023.12.07.570631v1