Multi-algorithm machine learning combined with in silico gene knockout reveals the diagnostic value and functional regulatory networks of ferroptosis-related genes in gastric cancer.

Ferroptosis plays a critical role in the occurrence and progression of gastric cancer (GC). This study aimed to systematically identify key ferroptosis-related genes (FRGs) in GC, construct a high-performance multi-gene model for early screening, and elucidate their potential functional networks. Based on the GSE184336 transcriptomic dataset, GC-related modules were screened using weighted gene co-expression network analysis (WGCNA). Differentially expressed genes, WGCNA key module genes, and reported FRG sets were intersected to identify candidate key genes. Innovatively, over 100 combinations of feature selection and machine learning algorithms were applied to construct diagnostic models, and their robustness was validated in two independent external cohorts. Single-cell transcriptomic data were used to analyze model gene expression, and in silico gene knockout analysis was performed to assess potential functional pathways. A total of 53 ferroptosis-related intersecting genes were identified. These genes were significantly enriched in ferroptosis and tumor-associated pathways. The optimal multi-gene diagnostic model (glmBoost + PLS) achieved an area under the curve (AUC) of 0.934 in the training set and 0.991 and 0.969 in two external validation cohorts, comprising five key genes: AKR1C1, CTSB, EZH2, IDO1, and TIMP1. Single-cell analysis revealed high expression of TIMP1, CTSB, and EZH2 in myeloid cells, fibroblasts, and immune cells. In silico knockout analysis demonstrated that these five model genes played critical roles in metabolic reprogramming, cell cycle regulation, and extracellular matrix modulation. By integrating multi-algorithm diagnostic modeling with in silico gene knockout analysis, this study systematically identified key FRGs and regulatory networks in GC.