|Home | About | Journals | Submit | Contact Us | Français|
Current algorithms for the calculation of peptide or protein pI, based solely on the charge associated with individual amino acids, can calculate pI values to within ±0.2pI units. Here we present a new pI calculation algorithm that takes into account the effect of adjacent amino acids on the pI value. The algorithm takes into account the effect of adjacent amino acids ±3 residues away from a charged aspartic or glutamic acid, as well as effects on the C-terminus, and applies a correction factor to the pK values of the charged amino acids. Large pK shifts are observed for the short-range interactions of aspartic and glutamic acid with the N-terminus and with internal histidine, lysine, or arginine residues. Conversely, interactions of aspartic and glutamic acid with hydrophobic, slightly polar, or other aspartic and glutamic acid residues are negligible. This results in a much narrower distribution of peptides across individual IPG-IEF fractions. The correction factors are derived from a 5000-peptide training set using a genetic optimization approach. The unique advantage of the genetic algorithms is that the pK optimization problem is independent of the evaluation function employed. This in turn allows the optimization function to determine the next set of solutions that can then be re-evaluated. This cycle continues until a set convergence criterion is reached. The accuracy of the new pI values obtained with this method approaches the error associated with the manufacture of the IPG strip (±0.05 pI units). The approach is demonstrated for cytosolic cell extracts derived from the breast cancer cell line DU4475, and from membrane preparations from human lung tissue samples.