Supplementary information.

<b>Fig A. Heatmaps of the ESPCA, SPCA, PCA model.</b> (A) the result of the ESPCA model. (B) the result of the SPCA model. (C) the result of the PCA model. <b>Fig B. Principal component score plots of UMAP, VAEs, ESPCA, SPCA, PCA model.</b> (A) the score plots of the UMAP model. (B) the score plots of the VAEs model. (C) the score plots of the ESPCA model. (D) the score plots of the SPCA model. (E) the score plots of the PCA model. <b>Fig C. Boxplots of Cu_75_FPKM, Cu_150_FPKM, Cu_300_FPKM, Cu_600_FPKM, Cu_1200_FPKM.</b> (A) the boxplots of the Cu_75_FPKM. (B) the boxplots of the Cu_150_FPKM. (C) the boxplots of the Cu_300_FPKM. (D) the boxplots of the Cu_600_FPKM. (E) the boxplots of the Cu_1200_FPKM. <b>Fig D. Heatmaps of the ESPCA, SPCA, PCA model.</b> (A) the result of the ESPCA model. (B) the result of the SPCA model. (C) the result of the PCA model. <b>Fig E. Principal component score plots of UMAP, VAEs, ESPCA, SPCA, PCA models.</b> (A) the result of the UMAP model. (B) the result of the VAEs model. (C) the result of the ESPCA model. (D) the result of the SPCA model. (E) the result of the PCA model. <b>Fig F. Boxplots comparing gene expression levels between P210 and T315I samples.</b> (A) the boxplots for P210 replicate 3. (B) the boxplots for T315I replicate 1. (C) the boxplots for T315I replicate 2. (D) the boxplots for T315I replicate 3. <b>Fig G. Noise probes correlation analysis and pathway number analysis plots.</b> (A) Distribution of P_value for probes included in the noisy probes. (B) Distribution of the number of individual gene probes associated with known pathways in the noisy gene probes. <b>Table A. The top two identified PC1 and PC2 loadings by SPCA and PCA. Table B. Time and memory usage comparison across methods.</b> (DOCX)