A large-scale evaluation of algorithms to calculate average nucleotide identity

Average nucleotide identity (ANI) is a category of computational analysis that can be used to define species boundaries of Archaea and Bacteria. Calculating ANI usually involves the fragmentation of genome sequences, followed by nucleotide sequence search, alignment, and identity calculation. The original algorithm to calculate ANI used the BLAST program as its search engine. An improved ANI algorithm, called OrthoANI, was developed to accommodate the concept of orthology. Here, we compared four algorithms to compute ANI, namely ANIb (ANI algorithm using BLAST), ANIm (ANI using MUMmer), OrthoANIb (OrthoANI using BLAST) and OrthoANIu (OrthoANI using USEARCH) using > 100,000 pairs of genomes with various genome sizes. By comparing values to the ANIb that is considered a standard, OrthoANIb and OrthoANIu exhibited good correlation in the whole range of ANI values. ANIm showed poor correlation for ANI of < 90%. ANIm and OrthoANIu runs faster than ANIb by an order of magnitude. When genomes that are larger than 7 Mbp were analysed, the run-times of ANIm and OrthoANIu were shorter than that of ANIb by 53- and 22-fold, respectively. In conclusion, ANI calculation can be greatly sped up by the OrthoANIu method without losing accuracy. A web-service that can be used to calculate OrthoANIu between a pair of genome sequences is available at http://www.ezbiocloud.net/tools/ani. For large-scale calculation and integration in bioinformatics pipelines, a standalone JAVA program is available for download at http://www.ezbiocloud.net/tools/orthoaniu.