Multi-Omics Data Analysis using Machine Learning for Cancer Prediction and Diagnosis

The Cyclin-Dependent Kinases are the core components coordinating the eukaryotic cell division cycle. Generally, the crystal structure of Cyclin-Dependent Kinases provides information on possible molecular mechanisms of ligand binding. In this regard, a novel method is introduced to find the subset of micro-RNAs responsible for cancer diagnosis and classification. The proposed method leverages on a perfect combination of standard tools in data mining to enforce prediction accuracy and select putative micro-RNA biomarkers. In particular, a support vector machine is used for classification, the component analysis is used for micro-RNAs selection, and a differential evolution algorithm is used to maximize accuracy. This study identified ten cancer -specific panels of micro-RNAs whose classification accuracy is higher than 92%. The next work is a large-scale analysis of next-generation sequencing data to identify biomarker signatures for several of the most common cancer types, i.e., bladder, colon, kidney, brain, liver, lung,
prostate, skin, and thyroid cancers. The primary purpose of this approach is to search putative gene biomarkers from a population of the healthy and tumor samples. The problem is mapped into the comparison of optimization algorithms.