Single-Cell and Machine Learning Integration Reveals Ferroptosis-Driven Immune Landscapes for Melanoma Stratification

Lei Wang¹Xue-ying Jin¹Yu-chen Wu²Ru-ning Qiu²Jianfang Wang^1*

• ¹Shaoxing People’s Hospital, Zhejiang, China
• ²Sir Run Run Shaw hospital, Zhejiang University, Zhejiang, China

Ferroptosis represents a critical death mechanism in melanoma progression and immune modulation. Here, we present a comprehensive multi-omics framework integrating bulk transcriptomics, single-cell RNA sequencing, and machine learning to decode the ferroptosis-immune axis in melanoma. We identified three ferroptosis-immune subtypes with distinct survival outcomes and immune phenotypes by applying consensus clustering and immune profiling across TCGA and GEO datasets. A 40-gene prognostic model, constructed through LASSO and stepwise Cox regression and validated externally, effectively stratified patients by survival risk and predicted sensitivity to chemotherapeutic agents. Single-cell transcriptomic analysis revealed elevated ferroptosis activity and immunosuppressive interactions, particularly involving POSTN-ITGB5 signaling between fibroblasts and immune cells. A clinically applicable nomogram integrating risk scores and clinical factors demonstrated robust survival prediction (AUC 0.829-0.845). A 4-gene prognostic model (CLN6, GMPR, AP1S2, ITGA6) was optimized via machine learning, and functional validation confirmed the role of CLN6 in proliferation and migration. This study provides a robust prognostic framework and therapeutic roadmap, enhancing precision immuno-oncology in melanoma.