oScience
Our Quantum AI technology features a complete and rigorous quantum mechanical representation of drug-target interactions. This is because biomolecular interactions are quantum mechanical in nature.
Bioactivity (ligand-target interactions) and pharmacological properties of a compound are defined by molecular electron densities – which are most accurately represented and modeled by quantum mechanics.
With Quantum AI, we have unique capabilities to identify novel active compounds and optimize simultaneously for multiple molecular properties.
Understanding Quantum AI
Our technology accurately describes and models biochemical structures and target interactions in quantum space
Quantum modeling of ligand-ligand similarity identifies chemotypically diverse molecules against target (with favorable IP profiles)
+Quantum analysis of ligand-target interaction using co-crystal data, allowing quantum “fingerprints” to be modeled
+Faster optimization of molecules in parallel (not series) facilitated by computational efficiency since pharmacological properties such as CNS uptake or toxicity are also best modeled in quantum space.
+oUnderstanding Quantum AI
Our technology accurately describes and models biochemical structures and interactions in quantum space (using quantum components).
Phenotypic quantum modeling of ligand-ligand interaction using bioassay data that identifies chemotypically diverse molecules against target (with favorable IP profiles)
+Quantum analysis of ligand-target interaction that focuses on mechanism of action using co-crystal data, allowing quantum “fingerprints” to be modeled
+
Faster optimization of molecules in parallel (not series) facilitated by computational efficiency since pharmacological properties – beyond bioactivity – such as toxicity and solubility are also best modeled in quantum space.
+
Because quantum wave functions are difficult to interpret in chemical space, we utilize:
This combination yields meaningful interpretations of many chemical concepts, such as energy partitioning, atomic softness, electronegativity equalization, atomic reactivity indices, as well as other molecular properties.
The product of this rigorous DFT / AIM theoretical framework is a map of well-defined, easily computable, localized molecular attributes, or quantum components. Quantum components present molecular electron flow and charge transfer properties in a form suitable for mathematical modeling.
For this, we introduce a quantitative metric (quantum similarity) to the chemical modeling space, from which metric modeling can leverage machine-learning, fuzzy-decision, AI algorithms. The result is molecular representation as a quantum object, which considers only a quantum representation of properties rather than chemical structure and traditional ligand-target interaction.
oLYN3009120 vs. Nilotinib
Working on BCR-ABL1 (an oncology target) to combat tumor resistance in drugs, we identified compound (LY3009120) that binds to allosteric pocket combinatorially with an “active site” inhibitor (Nilotinib).
Structural versus Quantum similarity between control inhibitor Nilotinib and in-silico discovered LY3009120. Structural diversity emanates from high quantum similarity (>0.9), which does not imply high chemical similarity (Tanimoto coefficient). Molecular surfaces are depicted in black/red, enveloping the color-coded atomic nuclei of the molecules. Molecular quantum components for BCR-ABL1 inhibition are projected onto the surface and shown by red regions.
oBCR-ABL1
Again, working on BCR-ABL1, our quantum analysis of co-crystal ligand-target structures does not require traditional docking methods and leads to greater structural flexibility, speed and accuracy.
.webp)
.webp)
oNrf2
For Nrf2 (our lead program in neurodegeneration) we identified a novel compound that activates Nrf2 and penetrates the blood-brain barrier (BBB).
Red area shows bioactive part of molecule in quantum space; i.e., the part that activates Nrf2.
Blue area shows part of molecule that allows for Blood Brain Barrier (BBB) penetration.
