Fig. 3: Results of the predictive model.
From: AI-guided few-shot inverse design of HDP-mimicking polymers against drug-resistant bacteria
a Comparison between predicted values and real measured values for β-amino acid polymers. Results show a desirable accuracy, with the metric of R-squared coefficient (R2) reaching 0.91, 0.88 and 0.91 on the values of the minimum inhibitory concentration for S. aureus (MICS.aureus), E. coli (MICE.coli) and the value of the minimum concentration to cause 10% hemolysis (HC10) on polymers with dimethyl (DM) subunit, and 0.92, 0.84 and 0.96 on polymers with monomethyl (MM) subunit. Text abbreviations (HE, OC, etc.) mean different hydrophobic subunits. All values are transformed values by natural logarithm. b–m Comparison between predicted values and real measured values for α-amino acid polymers (b), polymethacrylates (f) and polymethacrylamides (j). Predicted values on MICS.aureus (c, g, k) and MICE.coli (d, h, l) in various proportion are recorded with natural logarithmic transformation (log). We also visualize mean absolute error (MAE) of the predictions for each polymer (e, i, m) to show the difference when model transferring (Source data are provided as a Source Data file).
