Doctor of Philosophy (PhD)
Pathology & Microbiology
Hesham H. Ali Ph.D
Next Generation Sequencing (NGS) has emerged as a key technology leading to revolutionary breakthroughs in numerous biomedical research areas. These technologies produce millions to billions of short DNA reads that represent a small fraction of the original target DNA sequence. These short reads contain little information individually but are produced at a high coverage of the original sequence such that many reads overlap. Overlap relationships allow for the reads to be linearly ordered and merged by computational programs called assemblers into long stretches of contiguous sequence called contigs that can be used for research applications. Although the assembly of the reads produced by NGS remains a difficult task, it is the process of extracting useful knowledge from these relatively short sequences that has become one of the most exciting and challenging problems in Bioinformatics.
The assembly of short reads is an aggregative process where critical information is lost as reads are merged into contigs. In addition, the assembly process is treated as a black box, with generic assembler tools that do not adapt to input data set characteristics. Finally, as NGS data throughput continues to increase, there is an increasing need for smart parallel assembler implementations. In this dissertation, a new assembly approach called Focus is proposed. Unlike previous assemblers, Focus relies on a novel hybrid graph constructed from multiple graphs at different levels of granularity to represent the assembly problem, facilitating information capture and dynamic adjustment to input data set characteristics. This work is composed of four specific aims: 1) The implementation of a robust assembly and analysis tool built on the hybrid graph platform 2) The development and application of graph mining to extract biologically relevant features in NGS data sets 3) The integration of domain specific knowledge to improve the assembly and analysis process. 4) The construction of smart parallel computing approaches, including the application of energy-aware computing for NGS assembly and knowledge integration to improve algorithm performance.
In conclusion, this dissertation presents a complete parallel assembler called Focus that is capable of extracting biologically relevant features directly from its hybrid assembly graph.
Sommer, Julia, "Focus: A Graph Approach for Data-Mining and Domain-Specific Assembly of Next Generation Sequencing Data" (2017). Theses & Dissertations. 242.