CancerSPP

CancerSPP predicts skin cutaneous melanoma (SKCM) progression from genomic profiles to distinguish metastatic from primary tumors.

Key Features:

• Comprehensive genomic analysis: Utilizes mRNA, miRNA, and methylation data from The Cancer Genome Atlas (TCGA) to identify genomic features that differentiate metastatic and primary SKCM tumors.
• Machine learning models: Employs models including Support Vector Classification with Weight (SVC-W) using the expression of 17 mRNAs and reports evaluation metrics such as AUROC (0.95) and accuracy (89.47%) on independent validation datasets.
• Identification of key genomic features: Reports genes and miRNAs implicated in melanoma progression, including C7, MMP3, KRT14, LOC642587, CASP7, S100A7, hsa-mir-205, and hsa-mir-203b.
• Proposed novel biomarkers: Proposes putative metastasis-associated markers such as ESM1, NFATC3, C7orf4, CDK14, ZNF827, and ZSWIM7.

Scientific Applications:

• Therapeutic strategy optimization: Predictions of metastatic versus primary status can inform tailoring of therapeutic approaches for melanoma patients.
• Research and development: The identified genomic features and proposed biomarkers support mechanistic studies of melanoma metastasis and target discovery.
• Clinical decision support: Model-derived predictions and metrics can contribute evidence for prognosis and patient management decisions.

Methodology:

Analyzes TCGA SKCM mRNA, miRNA, and methylation data and develops machine learning classifiers including SVC-W using 17 mRNA features, with performance assessed by AUROC and accuracy on independent validation datasets.