Prediction of soluble heterologous protein expression levels in Escherichia coli from sequence-based features and its potential in biopharmaceutical process development

Prediction of soluble protein expression levels in Escherichia coli based on the nature of protein itself remains a challenge for bioprocess development (BD). This review will critically discuss the current efforts and achievements that employ computational approaches to develop prediction models for soluble protein expression in E. coli. The contrast between the remarkable progresses made on the predictive models achieved by bioinformatics and their relatively infrequent application in BD will be explained. The effects of process-relevant variables at four different levels on the expression of heterologous proteins, for example, gene, vector, host cell and cultivation process, and also a critical comparison of several established bioinformatics tools for predicting expression levels will be presented. The potential utility of this emergent technology to increase the efficiency of BD strategies and thereby to reduce the cost of establishing a process for soluble protein expression are critically examined.