Extended Data Fig. 10: Uncharacterized proteins are predicted to be broadly distributed and species-specific MF and CC terms with high confidence in the human microbiome.
From: Predicting functions of uncharacterized gene products from microbial communities
(a) Enumeration of the fraction of annotated taxa (first column), AUROC values per taxon (middle column; n = 3,849 total term-species pairs for prediction), and numbers of annotations (third column) for species-shared BP terms predicted by FUGAsseM in the HMP2. The top 15 terms with the largest number of species with at least one assignment are listed in decreasing order of average preserved annotations across species before running FUGAsseM (full results in Supplementary Table 15,16). ‘Before’ represents the number of annotations before running FUGAsseM. ‘After (default)’ represents the number after running FUGAsseM with the ‘default’ threshold, and ‘after (stringent)’ reflects values based on the ‘stringent’ threshold. Box plots display the median (line at the 50th percentile), interquartile range (box spanning the 25th to 75th percentiles), and whiskers (extending to 1.5× IQR). (b) Fraction of annotated taxa (first column), AUROC values per taxon (middle column; n = 89 total term-species pairs for prediction), and numbers of annotations (third column) for species-specific MF terms predicted by FUGAsseM from the HMP2 cohort. The 15 least frequently predicted terms are listed in decreasing order of mean number of preserved annotations as in a (full results in Supplementary Table 17,18). (c) Fraction of annotated taxa (first column), AUROC values per taxon (middle column; n = 926 total term-species pairs for prediction), and numbers of annotations (third column) for species-shared CC terms predicted by FUGAsseM in the HMP2. CC terms are listed in decreasing order of mean number of preserved annotations as in a (full results in Supplementary Table 15,16). Box plots as in a.
