Numerate leverages readily available data to address the key objectives of safety, efficacy and novelty in parallel, on the computer, at design-time. We deliver active, selective, synthesizable, and patentable leads.
The Drug Engineering Process
Step 1: Define drug design goals -- targets, potency, selectivity, toxicity, key patents, etc.
Step 2: Build lab-equivalent predictive models for each goal
Step 3: Apply models to search huge problem-specific virtual libraries
Step 4: Synthesize and verify biological properties in lab
Step 5: Optimize and identify candidate
We have made two key breakthroughs to enable this process:
- Accurately predicting the activity, ADME and toxicity of small molecule drugs.
We have invented cutting-edge machine learning algorithms to deal with the noise and biased nature of biological data to build models that are as accurate as laboratory assays.
Our models are ligand-based and therefore do not require the crystal structure of the target protein. Our models are also applicable to metabolism, PK, and certain safety-related issues. - Generating and searching very large chemical spaces.
For each project, we encode medicinal chemistry knowledge in the form of hundreds or thousands of core scaffolds and substitution groups representing 10 – 100 billion compounds.
We search these spaces using our predictive models to identify a few dozen compounds that are most likely to yield success in the laboratory.
Validation
The Numerate Drug Engineering Process has been extensively laboratory-validated and successfully applied to several therapeutic programs which demonstrate our ability to:
- Produce leads with nanomolar potency and in vivo efficacy
- Solve challenging selectivity issues
- Address complex multi-target profiles
- Invent novel, proprietary NCEs
- Work across therapeutic areas, target classes and types of modulation
Multi-Targeted Type 2 Diabetes Therapy
For a complete summary please see the case study.
We engineered leads that inhibit both the target of statins (HMG-CoA Reductase) as well as an enzyme central to inflammatory responses and energy balance (p38 MAP Kinase). This profile has the potential to address multiple aspects of cardiovascular risk associated with Type 2 diabetes.
Out of 19 compounds synthesized, eight met the multi-target specification. These represent four novel, distinct and patented chemical classes. Our leads display oral efficacy in vivo and differential effects on serum lipoprotein and triglyceride profiles relative to equal doses of Atorvastatin (Lipitor).
Versatile, Broad-Spectrum Medicine for HIV
We engineered novel non-nucleoside reverse transcriptase inhibitors (NNRTIs) that are active against multiple drug-resistant mutant strains, and possess improved pharmacokinetic and pharmaceutical properties.
Our leads have low nanomolar potency and broad-spectrum activity, along with improved solubility and plasma protein-binding relative to market-leading agents. We were able to meet the design goals in six months and with only 21 compounds synthesized and screened.
