Semi-supervised deep matrix factorization model for clustering multi-omics data

doi:10.1016/j.cmpb.2025.109094

Semi-supervised deep matrix factorization model for clustering multi-omics data

2025-10-08

0

PRE

AI

DOI：10.1016/j.cmpb.2025.109094

Original

Original request for help

Save

Abstract

En 中文

• This paper presents SSD-MO, a Semi-Supervised Deep Non-negative Matrix Factorization model designed to extract complex non-linear relationships from sparse, noisy, and high-dimensional multi-omics data. • SSD-MO features a semi-supervised component that leverages prior knowledge and optimal manifold assumptions to preserve the multi-omic data’s local geometry while generating a unique consensus representation for downstream clustering. • We conduct extensive experiments across multiple multi-omics datasets to evaluate clustering accuracy, normalized mutual information, and F-scores, complemented by analyses under different label ratios, as well as visualizations and context-specific interpretation of the proposed model.

Semi-supervised deep matrix factorization model for clustering multi-omics data

Abstract

Journal

Organization