CMF-Impute

CMF-Impute imputes dropout entries in single-cell RNA sequencing (scRNA-seq) expression matrices using collaborative matrix factorization to recover true gene expression and improve downstream analyses.

Key Features:

• Collaborative Matrix Factorization: Employs collaborative matrix factorization to impute missing entries by modeling associations between cells and genes.
• Performance Evaluation: Evaluated on six real-world scRNA-seq datasets and three simulated datasets, compared with five state-of-the-art methods, and assessed using sum of squared error and Pearson correlation coefficient for imputation accuracy and adjusted rand index and normalized mutual information for clustering using SC3 and t-SNE followed by K-means.
• Reconstruction of Biological Relationships: Reconstructs cell-to-cell and gene-to-gene correlations and supports lineage trajectory inference to recover biological interaction structures.

Scientific Applications:

• Cell Heterogeneity Analysis: Facilitates investigation of cell heterogeneity and identification of subpopulations within complex tissues by accurately imputing dropout events.
• Enhanced Data Interpretation: Improves accuracy of gene expression estimates to enable more precise biological interpretation of cellular functions and states.

Methodology:

Implemented as a MATLAB package that applies collaborative matrix factorization to model relationships between cells and genes for imputing technical dropouts in scRNA-seq data.