Journal: Frontiers in Chemistry

Article Title: SOMSpec as a General Purpose Validated Self-Organising Map Tool for Rapid Protein Secondary Structure Prediction From Infrared Absorbance Data

doi: 10.3389/fchem.2021.784625

Figure Lengend Snippet: (A) A 50-protein thin-film ATR reference set (see for list of proteins). Inset: Amide I maxima plotted versus total α -helix red and β -sheet blue content. Proteins F1−F7 (>60% helix) are purple; F8−14 (45–59% helix) are blue; F15−21 (34–44% helix) are turquoise; F22−F28 (26–33% helix) are green; F29−F33, F36, F38 (17–25% helix) are yellow; F34, F35, F37, F39, F40 (10–16% helix) are orange; F41−F50 (<10% helix) are red with the unfolded F50 dotted. (B) Overlay of some normalised ATR thin-film (solid lines) and aqueous transmission (dashed lines) spectra. (C) LOOV deviations of SS prediction from PDB structures for helix ( α -helix + 3 10 -helix) and β -sheet for the Amide I 50-protein thin-film reference set in order of decreasing helix content from left to right. 5 × NRMSD of the spectral fit is overlaid. Other category deviations are minus the sum of helix and β -sheet deviations. (D) Phosphoglycerate kinase (F17) LOOV SOMSpec output from 50-protein film reference set for a relatively poor quality example. In the map, U1, U2, U3 are the best matching nodes for the test protein. These can be expressed as linear combinations of their neighbouring reference set nodes. The proteins can be identified from in the SM, by noting that the test protein is F17 in the reference set, so proteins R1–R16 correspond to F1–F16, and R17–R49 correspond to F18–F50. The real spectrum is F17’s input data, and the predicted spectrum is the SOMSpec output.

Article Snippet: The SOMSpec App [coded in MATLabTM (MathWorks, Chatswood, Australia)] and example input and output files may be found in the data repository which can be accessed via the .

Techniques: Transmission Assay