Effect of Multi-K Contig Merging in de novo DNA Assembly

DNA Assembly is among the most fundamental and challenging problems in bioinformatics. Near optimal solutions are available for bacterial and small genomes. However assembling large and complex genomes including the human genome using Next-Generation-Sequencing (NGS) technologies is shown to be very difficult. This paper presents an algorithm for creating contigs from NGS short read data that is capable of working with multiple k-mer lengths and introduces a technique to combine contigs generated from different k runs with results from other assemblers in order to obtain significantly better assemblies. Experimental results from 9 real datasets show an increase in N50 value by a factor of 3, when combining newly created contigs with results from other assemblers.