Fig. 2

MIRTH achieves high accuracy imputing within datasets. a Samples for a subset of features were masked in half of all samples in a dataset before imputation to create data on which to assess imputation performance. b Imputation performance by dataset is reported by median \(\rho\) values across all simulated-missing features in each MIRTH iteration. c Imputation performance by metabolite, reported as the median \(\rho\) value for each metabolite across all trials, is plotted for each batch. Metabolites are ordered by decreasing imputation performance. d As dataset size (number of samples) increases along the x-axis, the proportion of well-predicted metabolites in a dataset increases as well. This illustrates the relationship between the number of training samples and overall imputation performance. e Imputation performance for each metabolite summarized across datasets (median \(\rho\) values across datasets are plotted). A subset of consistently well-imputed metabolites are labeled. Reproducibly well-predicted metabolites are indicated in blue. f The predicted ranks versus the true ranks of example metabolites, methionine and palmitate (16:0), when imputed in each single dataset. Each point represents one sample in which the metabolite was measured