Table 1 Performance metric assessment results from the Orthology benchmarking web service comparing conventional InParanoid analysis performed on fungal species with corresponding results from Domainoid and Inparanoid enriched with Domainoid. Results are from the Generalized species tree discordance benchmark. This benchmark relies on the axiom that trees of orthologs should reflect species phylogeny, such that disagreement between a trusted species tree and the reconstructed relationships acts as an inverse measure of quality of an orthology reconstruction method. For robustness this is assessed under multiple (usually 50 K iterations) resampling of sets of gene trees from the orthology inference results. From these, measures (average and standard error) of different agreement metrics are taken, with two basic metrics reported. One is Robinson-Foulds (RF) set (Jaccard) distance between splits encoded by species and gene trees (controlling for leaf set divergences). One is simply what fraction of reconstructed trees contain an error when compared to the trusted tree

From: Domainoid: domain-oriented orthology inference