Interpretable machine learning prioritizes transcriptomic features associated with ferroptosis-mitochondrial permeability transition net activity states in cervical cancer.
4 affiliations
- 1Department of Pain Medicine, Shaoxing People's Hospital, Shaoxing, China.
- 2Hangzhou Sun Taihe Traditional Chinese Medicine and Pharmacy, Hangzhou, China.
- 3Hangzhou Fuyang Hospital of TCM Orthopedics and Traumatology, Hangzhou, China.
- 4Reproductive Center, The Second Affiliated Hospital of Zhejiang Chinese Medical University, Hangzhou, China.
Abstract
Background
High-risk human papillomavirus (HPV)-associated cervical carcinogenesis involves oxidative stress, mitochondrial dysfunction, and altered non-apoptotic cell-death programs. Ferroptosis and mitochondrial permeability transition (MPT)-driven necrosis may share upstream stress-related transcriptional signals, but their joint pathway-state features in cervical cancer remain incompletely characterized.Methods
We constructed a directionality-aware net pathway activity framework using evidence-audited ferroptosis-promoting, ferroptosis-inhibitory, MPT-promoting, and MPT-inhibitory gene sets in 304 TCGA-CESC primary tumors. The score was benchmarked against ssGSEA, GSVA, and PLAGE and evaluated in controlled simulations. For feature prioritization, score-defining genes were excluded from predictors and Elastic Net feature selection, XGBoost tuning, and model fitting were performed strictly within repeated nested cross-validation. Outer-test SHAP, model-sensitivity analysis, and a 200-permutation complete-pipeline test were used for interpretation and robustness assessment. External validation used GSE138080, GSE63514, GSE223804, and GSE151666.Results
CombinedNet showed moderate-to-high concordance with ssGSEA, GSVA, and PLAGE (Spearman rho = 0.735, 0.882, and 0.943, respectively), while 1,300 controlled simulations showed strong recovery of the latent promoter-minus-inhibitor balance across scenarios and effect sizes (mean rho = 0.944; mean sign-recovery accuracy = 0.902; mean standardized RMSE = 0.305). The leakage-resistant repeated nested-CV model yielded AUC = 0.780 (95% CI 0.766-0.795); a matched complete-pipeline permutation test gave empirical p = 0.005. Stability-weighted SHAP and perturbation prioritized CKMT1B and KLK13, whereas GNA13, PANX2, and all five sterol genes were not selected in any of 25 outer-fold models. The sterol genes retained modest score correlations but showed no model sensitivity. Across three lesion-progression cohorts, no pathway score remained associated with stage after FDR correction; in the cancer-only GSE151666 RNA-seq cohort, CombinedNet also showed no FIGO-stage trend.Conclusion
This study supports directionality-aware transcriptomic scoring and a leakage-resistant feature-prioritization framework, but does not support GNA13/PANX2 or cholesterol-biosynthesis signals as stable model drivers or the composite score as a validated lesion-stage biomarker. The findings remain hypothesis-generating and require experimental validation.
Related
- An interpretable whole-blood transcriptomic model for Parkinson disease identification and molecular profiling
- The ferroptosis inhibitor NUPR1 coordinates the mitochondrial response to oxidative stress and cell metabolism during COPD pathogenesis in the lung
- Characterization of Ataxia-linked MFN2 Variants Reveals a Lipid Droplet/Mitochondrial Axis Driving Ferroptosis
- MGST1 expression reduces ferroptosis of macrophage in pediatric asthma through tricarboxylic acid-mitochondrial damage by ALOX5 ubiquitination.
- Integrated bulk and single-cell transcriptomic analysis of ferroptosis-related programs in high-grade serous ovarian carcinoma
BibTeX
@article{zhang2026interpretable,
title = {{Interpretable machine learning prioritizes transcriptomic features associated with ferroptosis-mitochondrial permeability transition net activity states in cervical cancer.}},
author = {Huiying Zhang and Yidong Zhu and Weijie Shao and Kai Xu},
journal = {Frontiers in genetics},
year = {2026},
month = oct,
doi = {10.3389/fgene.2026.1930345},
url = {https://europepmc.org/article/MED/42898594},
note = {CC BY},
}
RIS
TY - JOUR
TI - Interpretable machine learning prioritizes transcriptomic features associated with ferroptosis-mitochondrial permeability transition net activity states in cervical cancer.
AU - Huiying Zhang
AU - Yidong Zhu
AU - Weijie Shao
AU - Kai Xu
PY - 2026
DA - 2026/10/05
JO - Frontiers in genetics
DO - 10.3389/fgene.2026.1930345
UR - https://europepmc.org/article/MED/42898594
AB - Background High-risk human papillomavirus (HPV)-associated cervical carcinogenesis involves oxidative stress, mitochondrial dysfunction, and altered non-apoptotic cell-death programs. Ferroptosis and mitochondrial permeability transition (MPT)-driven necrosis may share upstream stress-related transcriptional signals, but their joint pathway-state features in cervical cancer remain incompletely characterized.Methods We constructed a directionality-aware net pathway activity framework using evidence-audited ferroptosis-promoting, ferroptosis-inhibitory, MPT-promoting, and MPT-inhibitory gene sets in 304 TCGA-CESC primary tumors. The score was benchmarked against ssGSEA, GSVA, and PLAGE and evaluated in controlled simulations. For feature prioritization, score-defining genes were excluded from predictors and Elastic Net feature selection, XGBoost tuning, and model fitting were performed strictly within repeated nested cross-validation. Outer-test SHAP, model-sensitivity analysis, and a 200-permutation complete-pipeline test were used for interpretation and robustness assessment. External validation used GSE138080, GSE63514, GSE223804, and GSE151666.Results CombinedNet showed moderate-to-high concordance with ssGSEA, GSVA, and PLAGE (Spearman rho = 0.735, 0.882, and 0.943, respectively), while 1,300 controlled simulations showed strong recovery of the latent promoter-minus-inhibitor balance across scenarios and effect sizes (mean rho = 0.944; mean sign-recovery accuracy = 0.902; mean standardized RMSE = 0.305). The leakage-resistant repeated nested-CV model yielded AUC = 0.780 (95% CI 0.766-0.795); a matched complete-pipeline permutation test gave empirical p = 0.005. Stability-weighted SHAP and perturbation prioritized CKMT1B and KLK13, whereas GNA13, PANX2, and all five sterol genes were not selected in any of 25 outer-fold models. The sterol genes retained modest score correlations but showed no model sensitivity. Across three lesion-progression cohorts, no pathway score remained associated with stage after FDR correction; in the cancer-only GSE151666 RNA-seq cohort, CombinedNet also showed no FIGO-stage trend.Conclusion This study supports directionality-aware transcriptomic scoring and a leakage-resistant feature-prioritization framework, but does not support GNA13/PANX2 or cholesterol-biosynthesis signals as stable model drivers or the composite score as a validated lesion-stage biomarker. The findings remain hypothesis-generating and require experimental validation.
SN - 1664-8021
ER -