Improving the portability and performance of jViz.RNA, a dynamic RNA visualization software

jViz.RNA is a Java based software that focuses on the visualization of RNA and its structural elements. It has been employed by many research groups around the world and has prompted excellent and constructive feedback from those groups, along with several suggestions for improvements. In this thesis, two major areas of jViz.RNA have been explored for the purpose of improvement; First, RNAML and FASTA file format support was added to jViz.RNA’s repertoire. This allows jViz.RNA users to utilize file formats used by other software, expanding jViz.RNA’s capabilities in working in pipeline systems, and also contributes to the standardization of RNA data exchange by supporting the use of RNAML. Second, five methods were explored in the context of improving the run time of jViz.RNA’s structure drawing algorithm. First, simple parameter optimization of the existing algorithm was attempted, then the use of the Barnes-Hut algorithm was explored, Thirdly, the effects of using multiple threads to handle the calculations were measured, additionally, the use of dynamic C libraries to integrate C code into the Java environment was investigated, and finally, a technique termed ’structure recall’, whereby the program uses files to register the layout of structures so they can be loaded for future runs, was examined. The results demonstrated that the use of approximation based techniques such as parameter optimization and the Barnes-Hut algorithm produces the most drastic improvements in run time, but does so at the cost of aesthetics, which may be unacceptable for visualization based software such as jViz.RNA. Multithreading and integration of C code, however, proved to be beneficial techniques since these improved the speed at which calculations are done, without distorting the structures in ways that obscure important information.