Abstract
Accurate and fast aligners are required to handle the steadily increasing volume of sequencing data. Here we present an approach allowing performant alignments of short reads (Illumina) as well as long reads (Pacific Bioscience, Ultralong Oxford Nanopore), while achieving high accuracy, based on a universal three-stage scheme. It is also suitable for the discovery of insertions and deletions that originate from structural variants. We comprehensively compare our approach to other state-of-the-art aligners in order to confirm its performance with respect to accuracy and runtime. As part of our algorithmic scheme, we introduce two line sweep-based techniques called "strip of consideration" and "seed harmonization". These techniques represent a replacement for chaining and do not rely on any specially tailored data structures. Additionally, we propose a refined form of seeding on the foundation of the FMD-index.