Optimising the altExtron gene splicing analysis pipeline

Exhibitor: Susan Lilley

Supervisor: Francis Clark

Research Group: Complex and Intelligent Systems

Industry Sector: Health / Medical Technology

Alternative splicing, the generation of multiple gene product isoforms from a single gene, is increasingly thought to be widespread within eukaryotic cells, making it a critical area of research in order to understand and apply the knowledge of a fundamental cellular process. The altExtron pipeline uses spliced alignment to computationally identify and analyse alternative splices for multiple organisms, using freely available gene and transcript sequence data. In its recent state, the pipeline was inefficient, both in terms of the computational time and the amount of human effort required for an update, and as a consequence the data sets were not updated as frequently as would be desired for effective research. This project performed three optimisations on the altExtron pipeline. First, since the data is largely unchanged between builds, the computational time required for an update was reduced by using the results from the previous build and only recomputing those results affected by the changed data. Second, since the speed of spliced alignment is heavily dependent upon how repetitive the sequence data are, the speed was improved by more effective masking of highly repetitious input sequences. Third, a series of shell scripts were created to partially automate the pipeline, reducing the human hours required for an update. The combination of these improvements has reduced the computational time taken to less than 5% and the number of required manual intervention points to less than 40% of the original. The work presented in this project allows research concerning alternative splicing to continue with a faster pipeline and therefore a more up-to-date data set.

Thesis Document (PDF)