Abstract
The self-assembly of low molecular weight gelators to form gels has enormous potential for cell culturing, optoelectronics, sensing, and for the preparation of structured materials. There is an enormous "chemical space" of gelators. Even within one class, functionalised dipeptides, there are many structures based on both natural and unnatural amino acids that can be proposed and there is a need for methods that can successfully predict the gelation propensity of such molecules. We have successfully developed computational models, based on experimental data, which are robust and are able to identify in silico dipeptide structures that can form gels. A virtual computational screen of 2025 dipeptide candidates identified 9 dipeptides that were synthesised and tested. Every one of the 9 dipeptides synthesised and tested were correctly predicted for their gelation properties. This approach and set of tools enables the "dipeptide space" to be searched effectively and efficiently in order to deliver novel gelator molecules.