For any kind of polyprotic system for instance DNA, RNA and protein, the isoelectric point can be expressed as the point in titration curve where the net charge of the ampholyte becomes zero. For defining the properties of various proteins and peptides, many biochemical methods rely on their isoelectric point. The 2-D gel electrophoresis also rely on the separation of proteins based on their isoelectric point and then elution and purification of isolated protein. Thus accurate theoretical prediction of pI would serve as an important point for such proteomic analysis.
Thus, Audain et. al. (2016) in their recent article have used the PIP database for accurate predictions of pI. While working with peptide mixtures they have found that machine learning algorithms especially SVM based ones show higher superiority. You can look at the algorithm and software that they have created at the following URL:https://github.com/ypriverol/pIR.
Reference:
Audain et al.(2016). Accurate estimation of isoelectric point of protein and peptide based on isoelectric point. Bioinformatics 32(6):821-827.