Fig. 3: Establishment of ProAtlas through HuPSA-based deconvolution.
From: Unveiling novel double-negative prostate cancer subtypes through single-cell RNA sequencing analysis
A Workflow illustrating the process of establishing ProAtlas using HuPSA-based deconvolution. B Heatmap displaying the deconvolution result. Each PCa sample (n = 877) was re-classified based on its molecular features defined by the HuPSA dataset. Samples were also annotated with NE score, AR score, correlation score with (prostatic) HOX code, and histology. C Principal component analysis (PCA) was conducted on the ProAtlas samples. AdPCa samples were further classified based on their Gleason score information. The dashed circles represent the 75% data ellipse of the corresponding group.
