Evaluating and Improving the Efficiency of Software and Algorithms for Sequence Data Analysis

With the ever-growing size of sequence data sets, data processing and analysis are an increasingly large portion of the time and money spent on nucleic acid sequencing projects. Correspondingly, the performance of the software and algorithms used to perform that analysis has a direct effect on the time and expense involved. Although the analytical methods are widely varied, certain types of software and algorithms are applicable to a number of areas. Targeting improvements to these common elements has the potential for wide reaching rewards. This dissertation research consisted of several projects to characterize and improve upon the efficiency of several common elements of sequence data analysis software and algorithms. The first project sought to improve the efficiency of the short read mapping process, as mapping is the most time consuming step in many data analysis pipelines. The result was a new short read mapping algorithm and software, demonstrated to be more computationally efficient than existing software and enabling more of the raw data to be utilized. While developing this software, it was discovered that a widely used bioinformatics software library introduced a great deal of inefficiency into the application. Given the potential impact of similar libraries to other applications, and because little research had been done to evaluate library efficiency, the second project evaluated the efficiency of seven of the most popular bioinformatics software libraries, written in C++, Java, Python, and Perl. This evaluation showed that two of libraries written in the most popular language, Java, were an order of magnitude slower and used more memory than expected based on the language in which they were implemented. The third and final project, therefore, was the development of a new general-purpose bioinformatics software library for Java. This library, known as BioMojo, incorporated a new design approach resulting in vastly improved efficiency. Assessing the performance of this new library using the benchmark methods developed for the second project showed that BioMojo outperformed all of the other libraries across all benchmark tasks, being up to 30 times more CPU efficient than existing Java libraries