Identification of Novel Chemically Selective Displacers Using Svm Classification Model and Parallel Batch Screening

A Support Vector Machine classification model was used in concert with batch parallel screening data to predict selective displacers for the separation of a binary protein mixture (ribonuclease A and alpha-chymotrypsinogen A) from a large commercially available molecule database. This new technique can quickly screen hundreds of potential molecules and identify desired molecules through computational power that significantly lowers the cost of experimental screening. Parallel batch screening and column displacement experiments were then used as a multi-stage experimental verification process, which serves to verify the selectivity of molecules identified through SVM classification and gives the whole screening protocol a balance of accuracy and speed. The result showed that one of the most promising selective displacers N'1'-(4-methyl-quinolin-2-yl)-ethane-1,2-diamine dinitrate identified by classification modeling displaced nearly all of the ribonuclease A from the binary protein mixture while displacing no alpha-chymotrypsinogen A, proved the existence of a chemical selective displacer. The features of this new class of displacers may enable new industrial protein purification techniques, such as chemically selective batch chromatography separations, to be developed in the future.