Related Experiment Video
Updated: May 23, 2026
07:41
A Robust Discovery Platform for the Identification of Novel Mediators of Melanoma Metastasis
Published on: March 8, 2022
A Mitoxyperilysis-Related Single-Cell and Machine-Learning Framework Defines an Immune-Cold Melanoma Phenotype and a
Yuze Zhou1, Jiahong Fang1, Lujing Fei1
1Department of Plastic Surgery, Xiangya Hospital, Central South University, Changsha, China, csu.edu.cn.
Human Mutation
|May 22, 2026
Summary
This study links mitoxyperilysis, a cell death process, to melanoma's tumor ecosystem. A new machine learning model predicts patient survival based on gene expression, identifying immune-cold tumors.
Area of Science:
- Immunology
- Metabolic pathways
- Cancer biology
Background:
- Mitoxyperilysis, a mitochondria-dependent cell lysis, is influenced by immune and metabolic signals.
- Its specific role and clinical significance in melanoma are not well understood.
Purpose of the Study:
- To investigate the clinical relevance of mitoxyperilysis in melanoma.
- To develop a prognostic model for melanoma patients using machine learning.
- To analyze the tumor microenvironment and cell-cell communication in relation to mitoxyperilysis.
Main Methods:
- Analysis of single-cell RNA sequencing data (GSE215120) to calculate a mitoxyperilysis-related score (MRS).
- Comparison of cell-cell communication in MRS-high versus MRS-low melanoma states.
- Machine learning (gradient boosting machine) applied to TCGA-SKCM and GEO cohorts for prognostic model development.
- Multialgorithm deconvolution and ESTIMATE used to assess the tumor microenvironment.
Main Results:
- The MRS score varied across melanoma cell types, indicating heterogeneity.
- MRS-high tumors showed increased and stronger intercellular communication, suggesting a rewired ecosystem.
- A gradient boosting machine (GBM)-based signature accurately stratified overall survival across multiple cohorts.
- High risk scores correlated with reduced immune/stromal signals, higher purity, and an immune-cold microenvironment.
- The gene GPR143, a representative model gene, was upregulated in melanoma, linked to worse survival, and its knockdown inhibited colony formation.
Conclusions:
- This study establishes an integrated framework connecting single-cell mitoxyperilysis programs with machine learning validation in melanoma.
- The findings enable mechanistic insights into immunometabolic heterogeneity.
- A validated prognostic signature offers clinically applicable risk stratification for melanoma patients.
