An Improved Method for Assembling Metagenomic Sequences

Abstract

Metagenomic assembly is a crucial step in analyzing pipelines of microbial communities. It allows to reconstruct contiguous consensus sequences (contigs) from mixed reads of multiple species. Meta-IDBA is one of the de novo assemblers which applies the de Bruijn graph algorithm to assemble metagenomic data without reference genomes. However, errors occurring after the graph creation are not well corrected, and thus result to low accuracy of the approach. This study proposes an improved method that tackles the problem using an effective error removal technique. Experimental results show that the improved algorithm outperforms original Meta-IDBA assembler on simulated and real data.