Super Molecule representing 1018 chemical entites
Platform & Technology
THE HQL PARADIGM
During the past two decades large efforts have been expanded towards moving the drug design process from a labor-intensive wet-lab approach to a computational-intensive, rational process. Although large numbers of algorithms and tools have been developed, these have not lived up to expectations to become the main engine of drug discovery. Despite this, computational tools are ubiquitous in the process.
In HQL we believe that a major factor contributing to this failure is the fact that essential, high resolution (3-dimensional) information is applied at late stages, on small numbers of compounds, since algorithms that can handle high resolution information are computationally highly demanding. Thus, there is a clear problem of sampling when applying accurate, high-information content computational tools. Additionally, it is well recognized that synergy of in-silico technologies is a key for substantial improvement.
HQL's ChemSpace Scanner (CSS) computational platform enables the synergy of multiple drug design algorithms and approaches, and enables screening of huge (order of magnitude >1030) chemical libraries against high-resolution, high information-content models.
The CSS platform is comprised of four main elements: (1) the digitized chemical libraries, (2) the screening algorithm itself, (3) a drug model against which the virtual library is screened, and (4) tools for scoring and ranking resulting hits.
- The chemical library: HQL developed a unique library representation, in which "super molecules", each represents orders of magnitude of 1011-1028 molecules and their 3-dimensional conformations are constructed. This is in contrast to all other virtual screening platforms in which require the enumeration of all of the molecules screened.
- The screening procedure: The ChemSpace Scanner Algorithms (CSSA), are composed of sets of filters that are applied in increasing order of complexity to the "super molecules"; this is in contrast to all available screening algorithms that are applied to discrete molecules. By screening "super molecules" rather than discrete molecules, each screening step screens trillions of compounds simultaneously. The result of applying the screening procedure to HQL's "super molecule" library is at least ten orders of magnitude more efficient than current technologies and therefore enables screening much larger numbers of molecules, using higher resolution models.
- The drug models: current drug models used for virtual screening come in various flavors, each with strengths and limitations, and it is currently recognized that the synergy of drug models is a key for substantial improvement. HQL’s comprehensive drug models (CDMs) combine information obtained from multiple state-of-the art computational technologies: pharmacophoric models and shape descriptors obtained from sets of active and inactive compounds, excluded volume information obtained from docking of active compounds into the target binding sites, chemical binding information obtained from fragment-based design methods, ADMET filters obtained from public and proprietary databases and tools, and empirical information obtained from medicinal chemistry expertise.
- Scoring and Ranking: HQL is incorporating state-of-the-art scoring functions into a proprietary clustering algorithm, ChemSpace-Cluster (CSC), which prioritizes the resulting leads. Once leads are identified, HQL uses state-of-the art medicinal and computational technologies to optimize them.
HQL'S TECHNOLOGICAL ADVANTAGE
- ChemSpace Scanner is able to screen chemical libraries on the orders of magnitude of 1030 entities against the high resolution, high-information content drug models, at the onset of a drug design project. In contrast, current technologies can screen order of magnitude of 108 - 1010 molecules at very limited resolution. High resolution models can be used to screen 3-4 orders of magnitude less.
- The CSS platform yields a full scope of diversity which is essential for the identification of the best drug candidates and high success rates. The outcome of applying CSS is a highly diverse set of chemical scaffolds that meet the requirements of the drug model. In contrast, current technologies have limited "scaffold hopping" capabilities yielding significantly more limited diversity.
- The ability to screen orders of magnitude of 1030 entities increases the chance of finding solutions in cases where these are rare.
- HQL's CDM enables the integration of inputs from multiple existing and foreseeable technologies.
HQL'S FUNCTIONAL ADVANTAGE
- The ability to screen orders of magnitude of 1030 entities enables dealing with larger, more flexible molecules than those tractable by current technologies. This ability is crucial for dealing with the design of small molecule protein-protein interaction inhibitors, one of the toughest challenges faced by the industry today.
- Multi-target drugs: New hypotheses are currently challenging the classical dogma that optimal drugs should be highly specific to a single target. The relatively new field of systems biology is revealing complex network structures that are often robust to point perturbations, and suggest that highly selective compounds may be clinically less effective than multi-target drugs. The integration of systems biology with clinical research has the potential to greatly expand the space of druggable targets, yet the rational design of compounds targeting several targets faces multiple challenges that require, among others, new computational methods for optimization of multiple drug models simultaneously. Due to the efficiency of HQL's virtual screening technology, multiple drug models can be screened in combination with structural information from the multiple targets.
- Fragment-based-drug-design (FBDD) is an emerging alternative to HTS/UHTS, and is gaining increased usage throughout the pharmaceutical industry. One of the major challenges face by this promising technology is the linking of fragments to obtain drug-like molecules. HQL’s platform obviates the need for the complex task of designing linkers by screening its huge “super-molecule” libraries for molecules that contain the bound fragments in the correct spatial orientation.