A THEORETICAL ANALYSIS OF SCALABILITY OF THE PARALLEL GENOME ASSEMBLY ALGORITHMS
Munib Ahmed, Ishfaq Ahmad, Samee U. Khan · 2014
A rapid growth of the sequenced genomic data over the last two decades has far exceeded the advancement of both the algorithms and the computing horsepower required to expeditiously process and analyze it. Several algorithms have been devised and implemented to assist the process of genome fragments assembly: one of the most challenging and computationally intensive processes that may take weeks to assemble large size genomes. A few such algorithms have also been parallelized to speed up the process. However, there is a need to analyze such parallel algorithms using the specific metrics of parallel computing to ascertain their scalability and efficiency. The fact that the problem size can vary from a few million units of data to several billions, along with the vast differences in the degree of repetition in data sets, calls for the ability to establish an association between the nature of the problem and the algorithm that best solves it. This paper analyzes the scalability of two most widely used parallel genome assembly algorithms using Isoefficiency (Grama, Gupta & Kumar 1993) metric which will help provide a guideline to determine when and how to choose a particular genome assembly technique based on the nature and the size of the problem being solved.