Beyond observation: genomic traits and machine learning algorithms for predicting fungal lifestyles

Figure S1. Distribution and proportion (%) of TE families in 638 genome assemblies. The bubble size represents the proportion of the TE in the genome. The bar represents the proportion of total TE size to the genome size. Figure S2. Composition and abundance of functional protein groups in 638 genome assemblies. The bubble size represents the number of the protein group. The bar represents the proportion of the secretome size to the total number of proteins per genome. Table S1. A summary table containing genome information of 638 genome assemblies, lineage information, and statistics of TE categories, basic genomic features and functional protein groups. Table S2. Taxonomic and lifestyle coverage of 638 Sordariomycete genomes. Table S3. General statistics of sequencing technologies, assembly completeness and TE sizes of 638 Sordariomycete genomes. Table S4. Pearson Correlations of 25 basic genomic features. Table S5. Comparative analysis results of 25 basic genomic features. Table S6. Pearson Correlations of 24 functional protein features. Table S7. Comparative analysis results of 24 functional protein features. Table S8. Predictive accuracies of six machine learning algorithms. Table S9. Predicted results of 85 undetermined genomes and the observed lifestyles of phylogenetically close groups.