Can Residual Thresholded Deep Matrix Factorization fix errors in big pharmacogenomic datasets?

Original title: Large-Scale Information Retrieval and Correction of Noisy Pharmacogenomic Datasets through Residual Thresholded Deep Matrix Factorization

Authors: Zhiyue Tom Hu,Yaodong Yu,Ruoqiao Chen,Shan-Ju Yeh,Bin Chen,Haiyan Huang

Pharmacogenomics has become a popular research area in precision medicine, but researchers have discovered a problem with drug sensitivity data. There is a significant inconsistency in the values among different datasets, which suggests that there is a lot of noise in the data. This noise makes it difficult to analyze and draw conclusions from the data. To address this issue, the authors of this article propose a new method called Residual Thresholded Deep Matrix Factorization (RT-DMF). This method uses deep learning to correct and impute the drug sensitivity data matrix. The authors explain that Deep Matrix Factorization (DMF) is a powerful technique for uncovering subtle patterns in data, and RT-DMF takes it a step further by incorporating an iterative residual thresholding procedure to retain signals that are likely to be therapeutically relevant. The effectiveness of the approach is demonstrated through validation using simulated and real pharmacogenomics datasets. The authors provide an open source package for implementing their method. This article is free of any competing interests.

Original article: https://www.biorxiv.org/content/10.1101/2023.12.07.570723v1