Non-parametric correction of estimated gene trees using TRACTION

Table 1 Empirical properties of the simulated datasets used in this study: gene tree heterogeneity, the average normalized RF distance between true gene trees and true species trees (GT-HET); average gene tree estimation error (GTEE); and the average distance of the ASTRID reference tree, to the true gene trees

	GT-HET	GTEE	Distance ASTRID to true gene trees
ILS-only, low ILS, 26 species [25]
# sites varies	0.10	0.32	0.08
ILS-only, high ILS, 26 species [25]
# sites varies	0.36	0.40	0.33
ILS+HGT, moderate HGT (m5), 51 species [29]
100 sites	0.54	0.63	0.55
250 sites	0.54	0.47	0.55
500 sites	0.54	0.47	0.54
ILS+HGT, high HGT (m6), 51 species [29]
100 sites	0.68	0.62	0.68
250 sites	0.68	0.46	0.68
500 sites	0.68	0.38	0.68

The publications from which the simulated datasets are taken are also indicated. In total we analyzed 68,000 genes with varying levels and causes of true gene tree heterogeneity (to the true species tree) and gene tree estimation error. The ILS-only conditions each had 20 replicates, and the ILS+HGT conditions each had 50 replicates

ISSN: 1748-7188