FlexNovo is a new molecular design program based on the FlexX-technology of incrementally constructing a ligand in an active site. It facilitates a fragment-based approach to de novo design in the context of the 3D-structure of the target.
- Fragment spaces consist of sets of molecular fragments that have specific 'link atoms' as open valences. Different 'link types' represent different chemical environments. A set of rules determines which link types are compatible (i.e. can be connected). Fragment spaces can be focused libraries derived from known binders or general fragment collections with link-rules. A space with only 16 000 fragments easily spans a chemical space of about 1018 virtual compounds of reasonable size, thus covering a considerable part of the 'chemical universe'.
- The FlexNovo algorithm is based on the incremental construction procedure as implemented in FlexX. In a first step FlexNovo places all available fragments. It takes the best scoring fragments entering the following steps during which compatible fragments are successively added to the so far obtained set of placements. All available scoring functions known from FlexX can be used for ranking the (intermediate) results. Additionally, FlexNovo considers a multitude of physicochemical property, geometry and diversity filter criteria. For the initial placement step, it also evaluates the compliance with pharmacophore-type constraints as known from FlexX-Pharm.
- In a first feasibility study, FlexNovo has been used to design potential inhibitors for four targets of pharmaceutical interest: dihydrofolate reductase, cyclin-dependant kinase 2, cyclooxygenase-2 and estrogen receptor. For each of these targets solution sets were generated that consist of up to 50 de novo molecules. All of the sets contained chemically reasonable structures that have similar structural motifs when compared to known active structures. The solutions also mimic the corresponding binding orientations. Additionally, the distribution of a number of physicochemical properties within such a set is reasonable and similar compared to distributions obtained from sets of known active structures.
FlexNovo is meant to be an integrated and user-customizable idea-generating system. It suggests new structural motifs, designing libraries of lead-prototypes and in proposing novel scaffolds as well as alternative binding modes for a target protein.
It also facilitates the evolution of a ligand in the depth of the active site, if one starts e.g. from a fragment obtained in a crystal structure screen.
The software can be even more powerful if there is expert knowledge integrated, for example, considering key interactions with the target, specific chemical motifs or certain physicochemical properties.
Fragment spaces as input for FlexNovo can be generated by CoLibri, BioSolveITs library design toolkit. Additionally, pre-defined fragment spaces derived from vendor compound collections can be obtained from BioSolveIT upon request. For similarity-based de novo design FTrees-FS is the method of choice.
FlexX's intuitive and easy to use GUI can be used to prepare the active site of the protein. This can then be readily imported into FlexNovo.