Extended Data Fig. 4: The factorization model based on machine learning and generalized linear model algorithm in cardiac snRNA-seq.

a, Schematic diagram of the generation of MlGlmCells factorization model that separates the dual effects of age and gender across different samples, which is based on machine learning and generalized linear model algorithm. Materials sourced from the Freepik website. b, Bar plot showing the prediction precision of age factorization for each cell type using the MlGlmCells model. TP: true positive (young samples that have been correctly identified), FN: false negative (young samples that have been misidentified), TN: true negative (aged samples that have been correctly identified), FP: false positive (aged samples that have been misidentified). c, Scatter plots showing the accuracy of age factorization performed by the MlGlmCells model for each cell type. d, Bar plot showing the prediction precision of gender factorization for each cell type using the MlGlmCells model. TP: true positive (male samples that have been correctly identified), FN: false negative (male samples that have been misidentified), TN: true negative (female samples that have been correctly identified), FP: false positive (female samples that have been misidentified). e, Scatter plots showing the accuracy of gender factorization performed by the MlGlmCells model for each cell type. f, Venn diagram showing the comparison of the number of differential genes analyzed by Seurat model and the MlGlmCells model. Panel a created with BioRender.com.