Kart adopts divide-and-conquer strategy. It finds an alignment by clustering all seeds on the reference genome. Kart divides a read into fragments that are easy to align (without INDELs) and those that require the dynamic programming algorithm, and align each fragment to compose the final alignment. Kart is the first method which use divide and conquer technique on the problem of NGS read mapping.

We simulated read libraries of the human genome and generates synthetic reads with random sequencing errors. We compared Kart with several existing nine popular read aligners, like BWA-MEM, Bowtie2 and HISAT2, etc. The experiment result showed that most selected aligners produced comparable alignments. However, Kart is the fastest aligner among all considered algorithms.

Our algorithm can not only handle NGS short read alignment and long read alignment, but it can also handle RNA-Seq read alignment, and genome sequence comparisons. It can be also applied to Metagenomics data analysis. Metagenomics data analysis involves huge amount of microbial sequences. We found that our algorithm can efficiently handle Metagenomics data and achieve strain-level resolution.