Fig. 2.

Species clustered according to their metabolic protein repertoires. A two-dimensional embedding of pairwise Hamming distances between presence/absence vectors of 2181 KEGG orthology (KO) identifiers was generated using the Uniform Manifold Approximation and Projection (UMAP) approach. The clades are color-coded according to the legend, and the total number of unique KO identifiers per species is coded by point size. The following species abbreviations are used: A.hoy, Azumiobodo hoyamushi; B.sal, Bodo saltans; E.gym, Eutreptiella gymnastica; E.gra, Euglena gracilis; H.pha, Hemistasia phaeocysticola; L.maj, Leishmania major; L.pyr, Leptomonas pyrrhocoris; N.des, Neobodo designis; N.gru, Naegleria gruberi; P.con, Paratrypanosoma confusum; Perk, Perkinsela sp.; Phyl. F, prokinetoplastid species PhF-6; Phyl. M, prokinetoplastid species PhM-4; R.cos, Rhabdomonas costata; R.hum, Rhynchopus humris; Sulc., Sulcionema specki; T.bor, Trypanoplasma borreli; T.bru, Trypanosoma brucei; T.gra, Trypanosoma grayi