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

Area of Science:

• Microbiology
• Bioinformatics
• Genomics

Background:

• Average nucleotide identity (ANI) is crucial for delineating species boundaries in Archaea and Bacteria.
• Traditional ANI calculation involves genome fragmentation, sequence searching, alignment, and identity computation, often using BLAST.
• The OrthoANI algorithm was developed to incorporate orthology into ANI calculations.

Purpose of the Study:

• To compare the accuracy and speed of four different algorithms for computing ANI: ANIb, ANIm, OrthoANIb, and OrthoANIu.
• To evaluate algorithm performance across a large dataset of over 100,000 genome pairs with varying sizes.
• To identify the most efficient and accurate method for ANI computation.

Main Methods:

• Comparison of four ANI algorithms: ANIb (BLAST), ANIm (MUMmer), OrthoANIb (BLAST), and OrthoANIu (USEARCH).
• Analysis of >100,000 genome pairs with diverse genome sizes.
• Performance evaluation based on correlation with ANIb and computational run-time.

Main Results:

• OrthoANIb and OrthoANIu showed strong correlation with ANIb across all ANI values.
• ANIm demonstrated poor correlation for ANI values below 90%.
• ANIm and OrthoANIu were significantly faster than ANIb, with run-time reductions of up to 53-fold for large genomes (>7 Mbp).

Conclusions:

• The OrthoANIu method offers a substantial speed-up for ANI calculations without compromising accuracy.
• OrthoANIu is a highly efficient tool for large-scale genomic analyses and bioinformatics pipelines.
• Web-service and standalone JAVA program for OrthoANIu are available for public use.