Meta-aligner: Long-read alignment based on genome statistics

Fast and accurate alignment of long-reads plays an important role in reducing the overall cost of long-read sequencing. In this paper, we propose Meta-aligner, an efficient and accurate long-read aligner that exploits the statistics of reference genome to improve performance in terms of reducing time complexity and achieving significantly higher recall for very noisy and long reads. The first step of algorithm adopts well-known short-read aligners in order to rapidly align a large fraction of reads through a progressive process of aligning read fragments to the reference genome. In the second phase, the remaining reads are handled by simultaneous alignment of all read fragments and a decision making process which exploits the overall information provided by the corresponding mapped fragments. By using this procedure, significant performance improvement is attained in comparison with traditional schemes in the case of PacBio long-reads.