Bradnam, K.R. et al. GigaScience 2, 10 (2013).

The Assemblathon 2 competition invited software developers to test their genome assemblers using sequencing reads from a bird, fish and snake genome. 21 teams competed, using data from any combination of three sequencing platforms. The organizers applied metrics such as scaffold and contig length, the presence of conserved genes, and accuracy compared to fosmid clones and optical maps to rank the programs. The snake genome had the highest-scoring assemblies, the fish the lowest, likely because of different levels of heterozygosity and repeat content. Overall the rankings as reported by Bradnam et al. showed that no software performed consistently well across all metrics within a species or across the same metric in different species. The take-home messages are that one should not rely on a single software tool for genome assembly and that it is misleading to judge the quality of an assembly based on only one metric.