Algorithmic and Computational Comparison of Metagenome Assemblers

Assembly of genome sequences of a microbial community is computationally challenging and complex than its single genome counterparts. Keeping in view the volume, diversity and varied abundance of different microbes, number of metagenome assemblers have been developed addressing specific associated computational issues mainly following De Bruijn Graph (DBG) and Overlap Layout Consensus (OLC) approaches. It is very pertinent to understand different computational approaches and issues of metagenomic assembly to further improve them with respect to time and computational resource requirements. Therefore, the main objective of this article is to discuss various metagenomics assemblers with respect to their development addressing major computational issues. Initially the computational perspective of single genome assemblers based on OLC and DBG graph construction approaches was described. This is followed by review of metagenomic assemblers with respect to the algorithm implemented for addressing issues in metagenome assembly. Further, performance of some of the popular metagenome assemblers were empirically evaluated with respect to their run time and memory requirements by taking diversified benchmark metagenomics data at ICAR-IASRI, New Delhi in 2019. It was concluded that performance of assemblers varied considerably on these datasets and there is further need to make an effort to develop new tools or to modify the existing ones using efficient algorithms and data structures.