Fig. 4: MetaGraph aids biological discovery.
From: Efficient and accurate search in petabase-scale sequence repositories
a, Associations between AMR gene families and RefSeq bacteriophages (phages) discovered in publicly available human gut microbiomes deposited to the NCBI SRA from 2008 to 2022. Squares with a black border indicate a statistically significant association and missing squares indicate hypotheses that were not tested due to an insufficient number of matching samples. inact., inactivation; prot., protein; protect., protection; res., resistant; ribo., ribosomal. b, The growth in the prevalence of resistance to antibiotics in human gut microbiomes across six continents from 2010 to 2020. Each point represents the normalized proportion of samples from a continent matching genes conferring resistance in a given year, and each line represents a binomial regression fit. Squares with a black border indicate regressions with significantly high fit scores and missing squares indicate experiments with an insufficient number of samples. CorrMCC, Matthews correlation coefficient. c, The relative fraction of GTEx (top) and TCGA (bottom) samples per tissue (x axis) that carry recurrent BSJs. The per-tissue ratio is computed as samples that carry at least 1 BSJ (dark blue) over all samples of that tissue (light blue)—both on the y axis. The absolute numbers are given in parentheses. d, The overlap between BSJ predicted from GENCODE and found through MetaGraph with the junction set experimentally provided through IsoCirc for GTEx (left) and TCGA (right) cohorts. Only junctions sufficiently highly expressed in the respective cohort are counted.
