Fig. 4

Taxonomic profiles characterization. a The 203 genomes frequencies in \(H^{(PG)}\) are displayed in pie-charts and clustered by successive taxonomic levels, i.e. taxa (outer ring), genus, class and phyla (inner ring), color-coded by phyla. Taxa names are displayed radially when their frequency is higher than 1% in the profile. These taxa are recapitulated in Table 3. b The same procedure is applied on MGS clustered at the genus level. Taxonomic levels are genus, class and phyla. c Average profile contribution in the 203 genomes counts. Namely, the same average profile weight \(\bar{W}^{(PG)}_{train}\) as in Fig. 3 is computed together with \(\bar{X}^{(PG)}_{train}\). Then, average profile contribution for genome j and profile i is computed with \(\bar{W}^{(PG)}_{train,i} H^{(PG)}_{ij} \left/ \bar{X}^{(PG)}_{train,j}\right.\). Finally, contributions are stacked by genome in bar plots and ordered by phyla. The residual \(1-\sum _{i=1}^4 \bar{W}^{(PG)}_{train,i} H^{(PG)}_{ij} \left/ \bar{X}^{(PG)}_{train,j}\right.\) is plotted in gray. Dotted gray lines indicate the value of \(\bar{X}^{(PG)}_{train,j}\) measuring the average AFT frequency (y log-scale on the right). d) The same procedure is repeated on the MGS clustered at the genus level