Fig 1. AGXT2 pseudospectrum and 3-aminoisobutyrate NMR spectrum, descriptions, and metabomatching match sets.

(A) The upper plot shows the experimental NMR spectrum of 3-aminoisobutyrate. The lower plot shows the (-log) p-values of the pseudospectrum of rs37369 in AGXT2, when p < 10⁻³. There is a close match between the experimental spectrum and the pseudospectrum, as the four sets of features that associate (p < 5 × 10⁻⁸) with rs37369 correspond to the principal peaks of the spectrum. (B) Taking a more detailed view of the spectrum descriptions (from HMDB), we see that the peaks of 3-aminoisobutyrate group into six clusters (labeled A through F). The multiplet ranges for clusters A, C, and E enclose their corresponding peaks well, padding by an average of 0.023 ppm. The multiplet range for cluster F is significantly wider, padding by 0.062 ppm. Approximating cluster areas as the product of the width of the cluster with the average height of the peaks in the cluster, then scaling, we find area-derived proton counts of 2.8, 0.7, 1.1, 1.2 for clusters A, C, E, and F, respectively. These counts are coherent with the listed proton counts of 3, 1, 1, and 1 for the respective multiplet ranges. Applying this same approximation for clusters B and D results in area-derived proton counts of 0.0, and 0.2. Because this implies corresponding multiplet proton counts of 0, we may consider the two spectrum descriptions essentially coherent, even though no multiplet ranges are listed for clusters B and D. (C) Match sets obtained from the peak and multiplet descriptions of the 3-aminoisobutyrate spectrum, for features resulting from a uniform NMR spectrum binning in 0.01 ppm increments, and with neighborhood parameter δ = 0.03 and 0.01, respectively. While the peak and multiplet descriptions of the 3-aminoisobutyrate NMR spectrum may be essentially coherent, their resulting match sets do differ, with 22 features unique to either one of the match sets.