Random forest analysis.

Feature importances derived from random forest classification of CBB-positive and CBB-negative genomes based on Enzyme Commission (EC) number or Pfam counts. The columns contain rank based on ’Importance’ (’Rank’), feature type (’Feature_Type’; DeepEC or Pfam), feature ID (’Feature’), feature name (’Name’), average feature importance for 100 random forests (’Importance’), coefficient of variation for feature importance based on 100 random forests (’CV_Importance’), feature description (’Description’; the DESC line from the Pfam HMM database, or the full list of enzyme names from KEGG for DeepEC), and the KEGG EC of the entry (’KEGG_EC’). Note that some DeepEC ECs were transferred to one or more new ECs in KEGG, as indicated by a discrepancy between ’Feature’ (if ’Feature_Type’ is DeepEC) and ’KEGG_EC’. A single feature can therefore be listed more than once. Also note that the importance and rank was calculated separately for DeepEC and Pfam feature types. (XLSX)