Fig. 2: Improvement in predictions using GutCP.
From: Ecology-guided prediction of cross-feeding interactions in the human gut microbiome
a Improvement in log error (\({\text{log}}_{10}(\frac{\,\text{pred}-\text{meas}}{\text{measurement}\,})\)) and the correlation between the prediction and measured fecal metabolome during 100 typical runs of the GutCP algorithm. The gray point at the top left indicates the performance of the original cross-feeding network of Ref. 9, and the black points at the bottom right, that of improved networks predicted using GutCP. A trajectory example, highlighting how performance improves over a GutCP run, is shown in red, and others are shown in gray. b Rarefaction curve showing the number of unique cross-feeding interactions discovered by GutCP over 100 runs of the algorithm. c Prevalence of links, i.e., the number of GutCP runs in which they repeatedly appeared (red dots; total 100 runs) and for comparison, a corresponding binomial distribution with the same mean (black dotted line). P values for different prevalences are estimated using the one-sided binomial test.
