Time efficient and novel ways of analyzing high-dimensional multi-omics datasets: parallel computing and multi-view learning

Abstract

Due to the advancements in high-throughput sequencing technologies high- dimensional omics data are rapidly increasing in number and require enhanced computational power to make sense. On the other hand, an integrated analysis of these omics data has the potential to reveal a comprehensive picture of any biological phenomenon than analyzing them separately (one omics data at a time). This thesis has two main contributions in the field of computer science and bioinformatics. First, the development of new CPU-based improved parallel algorithms for finding the LCS of DNA sequence data. Second, the thorough investigation of the existing canonical correlation analysis (CCA)-based multi-view learning methods for analyzing multi-omics data along with developing a new multi-view learning framework consisting of novel supervised DNN-based multi-view learning techniques.