酵素工学
オープンアクセス

概要

Purification Propensity for Proteins from Bacillus halodurans

Shaomin Yan and Guang Wu

The demand for proteins with special purposes increases significantly. These proteins are generally obtained through recombinant proteins, however their purification is costly and not easy. It is necessarily important to develop a method to estimate the chance of purification beforehand in order to have a prospective on proteins in question. Purification of a protein should be related to instinct properties of a protein including its 3D structure, and so far around 540 amino acid properties are found. Thus it is possible to test each amino acid property against the successful rate of protein purification to find out which property is more suitable to estimate the purification propensity. In this study, each of 535 properties was tested against 438 purified and 429 impossible purified proteins from Bacillus halodurans using logistic regression and neural network model. ROC analysis was applied to the resultant sensitivity and specificity. The results show that amino acid composition properties were generally less helpful to estimate the purification propensity whereas amino acid physicochemical properties, secondary structures and dynamic properties were more useful, and dynamic properties were more promising. Therefore several types of protein properties can serve to determine purification propensity of proteins, and have the potential to reduce the cost and to speed up the production in microbiological and biotechnical fields.