Fig. 3: Differentiating amino-acid biomarkers and a complex mixture using the αSA.

a Min-max scaled fluorescent signals from the αSA with tryptophan. Values are normalized relative to: 1, for the αHB and the reporter dye with no analyte; and 0, for the dye alone. Data shown corresponds to n = 9 independent samples. Boxes show the interquartile range with the median presented as a line. Whiskers show 1.5× interquartile range, or the range if a smaller value. Outliers are shown as diamonds. b Principal component analysis of the 5 amino acids: glutamate, blue circles; arginine, green triangles; serine, red squares; valine, cyan diamonds; and tryptophan, purple stars. c Representative dye-displacement data for each analyte in the AA group. αHB ID is shown above each fingerprint. In these cases, min-max scaled dye displacement is colored from dark red (less displacement) to dark blue (more displacement) according to the respective heat maps (right-hand side of each panel). Each fingerprint corresponds to the median signal across all repeats for each AA. d The 10 features selected to take forward to classification. Color scheme as in c, αHBs not selected are colored gray. e, Confusion matrix generated from the classification of AA samples using the Gaussian Naïve Bayes algorithm with nested stratified cross-validation. Here the coloring scheme is from dark red (all prediction) to dark blue (no predictions) according to the heat map (right-hand side). f Min–max scaled fluorescent signals from the αSAs challenged with different teas. Values are normalized as in (a). Black tea, black bars; green tea, green bars; Earl Grey tea, gray bars. Data shows corresponds to 178 independent samples (n = 59 black, n = 59 green and n = 60 Earl Grey). Box and whiskers are presented as in a. g Principal component analysis of the 178 brewed tea samples: black teas, black circles; green teas, green triangles; Earl Grey teas, gray squares. h Representative dye-displacement data for select tea samples (full range shown in Supplementary Figure 18). αHB ID is shown above each fingerprint. Color scheme as in (c). In this case, each fingerprint corresponds to the median signal of the 6 independent tea samples for each brand of tea. i The 4 features selected to take into classification. Color scheme as in (c), αHBs not selected are colored gray. For visualization purposes, the fingerprints in h and i are the median fingerprints from the 6 independent repeats for each tea brand rather than the 178 individual fingerprints. j Spearman coefficients of the αHBs in the αSA for the tea fingerprints. Color scheme is from strong correlation (dark red) to no correlation (dark blue) according to the heat map (right-hand side). k Confusion matrix generated from predictions of tea samples using the SVC algorithm with nested stratified cross-validation. Color scheme as in (e). Source data are provided as a Source Data file.