Figure 1. Average nucleotide identity from BLAST (ANIb) as a function of branch length scale factor α_BL.

Sampled on a log-scale, the parametric sweep crosses the operational species definition (95% ANIb) roughly midway (dashed grey horizontal line). A 95% similarity threshold is also used internally within IDBA-UD assembler (Peng et al., 2012) to determine whether to merge highly similar contigs and has been proposed as a pragmatic definition of bacterial species (Konstantinidis, Ramette & Tiedje, 2006; Richter & Rosselló-Móra, 2009) akin to 97% 16S rRNA identity.