Evaluation of nine popular de novo assemblers in microbial genome assembly

- Nine state of the art assemblers were evaluated and compared, in terms of accuracy and contiguity.
- Effects of different read coverages and length of k-mers were also evaluated.
- SPAdes and IDBA-UD were significantly better than others.

Next generation sequencing (NGS) technologies are revolutionizing biology, with Illumina being the most popular NGS platform. Short read assembly is a critical part of most genome studies using NGS. Hence, in this study, the performance of nine well-known assemblers was evaluated in the assembly of seven different microbial genomes. Effect of different read coverage and k-mer parameters on the quality of the assembly were also evaluated on both simulated and actual read datasets. Our results show that the performance of assemblers on real and simulated datasets could be significantly different, mainly because of coverage bias. According to outputs on actual read datasets, for all studied read coverages (of 7 ×, 25 × and 100 ×), SPAdes and IDBA-UD clearly outperformed other assemblers based on NGA50 and accuracy metrics. Velvet is the most conservative assembler with the lowest NGA50 and error rate.