PANoptosis‑Related Pathomics Prognostic Model for ovarian cancer (OC).
This repository accompanies our open‑access publication in Journal of Cellular and Molecular Medicine (2025).
Paper: Zhang Y, Fang M, Wang X, et al. Development of a PANoptosis‑Related Pathomics Prognostic Model in Ovarian Cancer: A Multi‑Omics Study. J Cell Mol Med. 2025;29:e70958. https://doi.org/10.1111/jcmm.70958
PANPM integrates hand‑crafted pathomics features (CellProfiler) and deep features (ResNet‑50) from H&E whole‑slide images, links them to PANoptosis activity, and builds a prognostic risk score model validated across external cohorts.
Our original goal was to explore a practical connection between pathomics signals in routine H&E slides and underlying biological mechanisms, focusing on PANoptosis as a proof‑of‑concept. While the modeling framework is intentionally simple and reproducible, the project provided valuable experience for our group in end‑to‑end WSI processing, feature engineering, and mechanism‑oriented interpretation—informing and accelerating our subsequent work.
- Code for WSI preprocessing/tiling and feature extraction
- Model training and evaluation (C‑index/AUC, KM, nomogram)
- Reproducible scripts/notebooks for key figures
This work uses public resources including TCGA/GTEx, GEO (GSE184880), and publicly available pathology/validation cohorts (e.g., TCIA/PLCO as described in the paper).
If you use PANPM, please cite:
Zhang Y, Fang M, Wang X, et al. J Cell Mol Med. 2025;29:e70958. doi:10.1111/jcmm.70958
Issues and PRs are welcome.