Abstract P32: Integrative Transcriptomic and Deep Learning Model Identifies Anti-Cancer Peptides Targeting KIF2C in Breast Cancer

Abstract

Breast cancer is the most diagnosed cancer and a leading cause of cancer-related death among women worldwide. Drug resistance is a major challenge, contributing to treatment failure and disease progression. Anti-cancer peptides (ACPs), small protein fragments, offer a promising alternative due to their selectivity, low toxicity, and potential for targeted therapy. In this study, we applied integrative bioinformatics and deep learning–based peptide screening to identify novel therapeutic targets and ACPs for breast cancer. Analysis of The Cancer Genome Atlas (TCGA) breast cancer dataset revealed 1,417 differentially expressed genes enriched in neuroactive ligand–receptor interaction and cytokine–cytokine receptor pathways. Notably, KIF2C emerged as a key upregulated hub gene associated with poor prognosis, suggesting its potential as a therapeutic target. Peptides were generated based on the amino acid composition of known ACPs and screened using machine learning and deep learning models. Top candidates were then evaluated for toxicity, allergenicity, and cell-penetrating ability, identifying peptides with a high probability of ACP activity and low toxicity. Among the nine top candidates, csACP-285 showed the strongest binding to the KIF2C motor domain via molecular docking, supported by favorable binding free energy (ΔG) and stable hydrogen bonding. These findings highlight KIF2C as a viable therapeutic target and csACP-285 as a promising ACP candidate for targeted, low-toxicity breast cancer therapy.