The Weekend Warrior: How to Build a Genomic Supercomputer in Your Spare Time Using Streams and Actors in Scala
Brendan Lawlor, Paul M Walsh · 2016
The cost of developing parallelised software is a significant bottleneck in the implementation of bioinformatics pipelines. This paper identifies the use of low-level threading abstractions as one cause for these elevated costs,, proposes alternatives which offer better value. We identify a core set of higher abstractions to accomplish low-cost multi-core, multi-server parallelization - Streams, Actors, the Scala language (StAcS). We present the performance results of a robust, parallelizable, scalable implementation of the Smith-Waterman algorithm developed using the StAcS approach at development costs inferior to equivalent systems written using current standard techniques.