Fig. 3

Validation experiment. A Validation experiment’s schematic figure. Step A―Predicting the recipient’s property post-FMT. Inserting all the MIPMLP preprocessed donors from all the cohorts (including donors that had not been actually transplanted) into the pre-trained iMic model, using existing datasets. iMic returns the predicted recipients’ properties post-FMT. Step B―Grouping samples for FMT validation experiments. Two groups were defined from the predicted properties. A group of predicted to be high (with predicted high values of the property) and predicted to be low (with predicted low values of the property). Step C―FMT experiment timeline. Two groups of mice were raised till the age of 6 weeks. They got antibiotic treatment (weeks 6–8). At the age of 8 weeks, stools were collected and sequenced. At the age of 8 weeks + 1 day, they received the first FMT. One group of mice got the FMT from the predicted to be high donors group, and the second group of mice got the FMT from the predicted to be low group. They got the second FMT from the same donors at the age of 9 weeks. Stool samples were collected a week after the second transplant at the age of 10 weeks and 6 weeks after the second FMT treatment at the age of 15 weeks. Step D―Recipient samples analysis. The real recipients’ properties were calculated from the mice’s stool samples at the age of 10 weeks. A comparison between the 2 groups’ properties showed significant differences in the targeted properties. B, C  Difference between Shannon diversities (B) and Euclidean distance between two recipients preprocessed ASVs (order-level) vectors (C). The right bar represents the distances between samples of different recipients (ABX mice) that got FMT from the same donor, SDDR, and the left bar represents the distances between samples of different recipients (ABX mice) that received FMT from different donors, DDDR. There is a significant difference between same and different donors post-FMT (10W), but not before (8W) (\(*\) \(p<0.05\), \(**\) \(p<0.01\), \(***\) \(p<0.001\)). D–G Differences between the donors and real recipients at different time points (8W, 10W) properties on the groups we defined, Shannon (D), Bacteroidales order relative abundances (E), Desulfovibrionales order relative abundances (F), and Verrucomicrobiales order relative abundances (G). The (A) represents a transplant from an adult human, while (Y) represents a transplant from a young human. Again, there is no difference before the FMT (8W) or in the donors (Donor)