Whole-Genome Assembly: An Experimental Study of Computational Costs and Architectural Opportunities

Abstract

Whole-genome sequencing (WGS) pro- vides a huge amount of reads from which a comple- te genome could be assembled. The recent advent of long read sequencing technologies, such as PacBio and Oxford Nanopore, and the subsequent appearance of high quality long reads (single molecule high-fidelity, or HiFi) have improved the scaffolding of the genome. However, both biology and computing communities still face great challenges in terms of computational cost. Thus, it is essential a high precision characte- rization of the methods for a correct identification of the main computing bottlenecks. This study will allow us to design new methods to mitigate compu- tational costs without losing accuracy and to adapt such methods to fully exploit new architectures that provide support to handle big amounts of data. In this paper, we experimentally study and characterize the most used whole-genome assemblers in order to design new approaches in this field.