Over the past decade, the evolution of next-generation sequencing technology has considerably advanced the genomics research. As a consequence, fast and accurate computational methods are needed for analyzing the large data in different applications. The research presented in this dissertation focuses on three areas: RNA-seq read mapping, large-scale data query, and metagenomics sequence classification. A critical step of RNA-seq data analysis is to map the RNA-seq reads onto a reference genome. This dissertation presents a novel splice alignment tool, MapSplice3. It achieves high read alignment and base mapping yields and is able to detect splice junctions, gene fusions, and circular RNAs comprehensively at the same time. Based on MapSplice3, we further extend a novel lightweight approach called iMapSplice that enables personalized mRNA transcriptional profiling. As huge amount of RNA-seq has been shared through public datasets, it provides invaluable resources for researchers to test hypotheses by reusing existing datasets. To meet the needs of efficiently querying large-scale sequencing data, a novel method, called SeqOthello, has been developed. It is able to efficiently query sequence k-mers against large-scale datasets and finally determines the existence of the given sequence. Metagenomics studies often generate tens of millions of reads to capture the presence of microbial organisms. Thus efficient and accurate algorithms are in high demand. In this dissertation, we introduce MetaOthello, a probabilistic hashing classifier for metagenomic sequences. It supports efficient query of a taxon using its k-mer signatures.

Technological advances in next-generation sequencing and biomedical imaging have led to a rapid increase in biomedical data dimension and acquisition rate, which is challenging the conventional data analysis strategies. Modern machine learning techniques promise to leverage large data sets for finding hidden patterns within them, and for making accurate predictions. This dissertation aims to design novel machine learning-based models to transform biomedical big data into valuable biological insights. The research presented in this dissertation focuses on three bioinformatics domains: splice junction classification, gene regulatory network reconstruction, and lesion detection in mammograms. A critical step in defining gene structures and mRNA transcript variants is to accurately identify splice junctions. In the first work, we built the first deep learning-based splice junction classifier, DeepSplice. It outperforms the state-of-the-art classification tools in terms of both classification accuracy and computational efficiency. To uncover transcription factors governing metabolic reprogramming in non-small-cell lung cancer patients, we developed TFmeta, a machine learning approach to reconstruct relationships between transcription factors and their target genes in the second work. Our approach achieves the best performance on benchmark data sets. In the third work, we designed deep learning-based architectures to perform lesion detection in both 2D and 3D whole mammogram images.