طراحي پردازشگر مبتني بر دِنا جهت تشخيص زيستنشانگرهاي موثر در سرطان

Breast cancer, microRNA, Onco-mir, DNA-based computing, molecular sensor, biomarker selec‎tion

Designing a DNA-based processor to detect biomarkers effective in cancer

مهندسي معماري كامپيوتر

Advances in early an‎d accurate disease detection, particularly in cancer, play a crucial role in improving treatment outcomes; however, current clinical methods are often costly, time-consuming, an‎d prone to error. In this study, with the aim of developing a rapid, low-cost, an‎d accessible diagnostic system for breast cancer, gene-expression data retrieved from the GDC database were analyzed using bioinformatics approaches, including Differential Gene Expression (DEG) analysis an‎d Weighted Gene Co-expression Network Analysis (WGCNA). These analyses led to the identification of 12 microRNAs with key roles in breast cancer. Based on the associated gene modules an‎d diagnostic relevance of these microRNAs, logical relationships between up-regulated an‎d down-regulated microRNAs were derived. To implement this logic, three DNA-based computational processor architectures were proposed: (1) a processor employing sequential logic gates an‎d two detection units (up-regulation an‎d down-regulation), (2) a redesigned structure using modified self-assembling tiles, an‎d (3) a processor built upon three-input Majority gates, enabling enhanced cascading capability, scalability, an‎d simultaneous multiinput detection. A portion of the first design was fabricated at the Gene Therapy Center of Shariati Hospital, where its performance was experimentally validated. Simulation results demonstrated that all three designs achieved PPV ≈ 0.80 an‎d NPV ≈ 0.99, which was expected due to their shared logical framework. Using a gradient boosting technique, an F1-score of 0.99 was obtained. Furthermore, 10- fold cross-validation yielded PPV = 0.99 an‎d NPV = 0.897, confirming the high accuracy an‎d stability of the system without biased inclination toward false detection. Overall, the findings highlight the efficiency an‎d high precision of the proposed approach for early breast cancer diagnosis an‎d its strong potential as a low-cost, scalable molecular sensing platform.

5

156432