We have developed a predictor of protein pKa shifts that employs a coarse-grained lattice presentation of protein structure. This software is called PKA17. Values of the parameters used in our software are produced by fitting to experimental acidity constants of Asp, Glu, Lys, and His. The program was tested upon the Song et al. 1 database which consisted of 34 different proteins with 284 experimentally know amino-acid pKa values. The calculations yielded an overall root-mean-squared deviation (RMSD) of 0.70 pH units. The RMSD values of the specific residues being 0.73 for Asp, 0.72 for Glu, 0.61 for Lys, and 0.71 for His. The PKA17 software was also tested again the 80 residues in the Stanton and Houk 2 dataset which contained significantly varying pKa deviations of pKa values from those of the isolated amino-acids in aqueous solution. (ΔpKa ≥ 1). The RMSD values for our technique with respect to this test set were 0.93 pH units overall, 0.83 for Asp, 0.74 for Glu, 1.0 for Lys, and 1.1 for His. The performance of PKA17 is at least on par with other popular predictors (such as PROPKA 3) and exceeds it in some important respects.

Coarse-grained predictor of protein pKa values

255th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.