Mapreduce in computational biology - A synopsis

In the past 20 years, the Life Sciences have witnessed a paradigm shift in the way research is performed. Indeed, the computational part of biological and clinical studies has become central or is becoming so. Correspondingly, the amount of data that one needs to process, compare and analyze, has experienced an exponential growth. As a consequence, High Performance Computing (HPC, for short) is being used intensively, in particular in terms of multi-core architectures. However, recently and thanks to the advances in the processing of other scientific and commercial data, Distributed Computing is also being considered for Bioinformatics applications. In particular, the MapReduce paradigm, together with the main middleware supporting it, i.e., Hadoop and Spark, is becoming increasingly popular. Here we provide a short review in which the state of the art of MapReduce bioinformatics applications is presented, together with a qualitative evaluation of each of the software systems that have been here included. In order to make the paper self-contained, computer architectural and middleware issues are also briefly presented.