Fig. 6: Prospective experimental validation against human receptors in vitro.

Using a cheminformatic protein target prediction approach (SEA^25,26) with the Twin-NN phenotypic distance, we make the mechanism of action predictions for NT-650 compounds and test them experimentally by radioligand binding assays. a Left: Top 5 DIVERSet compounds (represented by their motion index time-series, rows 2–6, gold) matched by Twin-NN distance to the NT-650 drug Imetit (top row, purple). Right: Diagram of this “phenoblast” approach. NT-650 drugs are columns. DIVERSet compounds are rows, ordered by Twin-NN phenotypic distance. Supplemental Table 1 maps compound IDs to supplier IDs. All time series in this plot are scaled to the minimum and maximum of the dataset (0 and 6750 MI units, respectively), and the y-axis is plotted on this normalized 0 to 1 scale. b Primary radioligand binding assays (binding inhibition at 10 µM, %) for 7 known drugs. The heatmap shows 5 DIVERSet compounds selected for testing (rows), with the SEA-predicted human protein targets as columns. c Same as (b) for secondary assays (dose-response radioligand binding experiments). d Representative dose-response curves from (c) for selected DIVERSet compounds tested against two human targets: the histamine H3 receptor (left) and the 5-hydroxytryptamine 2B receptor (right). Results (mean \(\pm \,\)SEM) from a minimum of 3 independent assays (each in triplicate) were normalized, pooled, and fitted to the built-in one-site competition binding function in the GraphPad Prism V10. Source data are provided in the Source Data File.