Machine learning based methods of analysing drug-like molecules
There is provided a method for a machine learning based method of analysing drug-like molecules by representing the molecular quantum states of each drug-like molecule as a quantum graph, and then feeding that quantum graph as an input to a machine learning system.
1. A machine learning based method of modelling a thermodynamic ensemble or representation of a drug-like molecule, in which a sample of the thermodynamic ensemble or representation is synthetically generated and inputted into a machine learning system, the thermodynamic ensemble or representation being a molecular orbital representation or a quantum graph representation of the drug-like molecule,
and in which the quantum graph representation is a molecular graph representation in which each node corresponds to a molecular orbital and edges correspond to a type of quantum correlation between molecular orbitals.
2. The machine learning based method of claim 1 , in which every element of the thermodynamic ensemble or representation can be represented as a Q-graph or tensor network, or the molecular orbital representation.
3. The machine learning based method of claim 1 , in which the quantum graph representation is the molecular graph representation of the drug-like molecule obtained from quantum mechanical calculations.
4. The machine learning based method of claim 1 , in which the quantum graph representation depends on a conformational state of the drug-like molecule.
5. The machine learning based method of claim 1 , in which the molecular orbital representation is a tensor network representation of molecular quantum states of the drug-like molecule.
6. The machine learning based method of claim 1 , in which synthetic generation of one or more samples of thermodynamic ensembles or representations is based on a thermodynamic quantity.
7. The machine learning based method of claim 1 , in which the machine learning system is configured to output a thermodynamic quantity based on an approximate expectation value over an entire or a representative set of the thermodynamic ensemble or representation.
8. The machine learning based method of claim 1 , in which the machine learning system is configured to learn a distribution of Boltzmann weights of an entire or a representative set of the thermodynamic ensemble or representation of the drug-like molecule.
9. The machine learning based method of claim 1 , in which determining a cost function or backpropagation of the machine learning system is based on a thermodynamic quantity to be outputted by the machine learning system.
10. The machine learning based method of claim 1 , in which a size of the synthetically generated sample of the thermodynamic ensemble or representation is tuned depending on a downstream application.
11. The machine learning based method of claim 1 , in which the machine learning system is a graph convolutional neural network.
12. The machine learning based method of claim 1 , in which the synthetically generated sample of the thermodynamic ensemble or representation is inputted as a molecular graph.
13. The machine learning based method of claim 1 , in which the machine learning system is configured to output any quantity that is a function of the thermodynamic ensemble or representation.
14. The machine learning based method of claim 1 , in which the machine learning system is used to predict ligand protein binding affinity, and in which synthetically generated samples of the thermodynamic ensemble or representation of a ligand in solution, of a protein in the solution, and of a ligand-protein complex are inputted into the machine learning system.
15. The machine learning based method of claim 1 , in which the machine learning system is used to predict ligand protein inhibition concentration, and in which synthetically generated samples of the thermodynamic ensemble or representation of a ligand in solution and of a ligand-protein complex are inputted into the machine learning system.
16. The machine learning based method of claim 1 , in which the machine learning system is used to predict lipophilicity of the drug-like molecule, and in which synthetically generated samples of the thermodynamic ensemble or representation of a unionized and/or ionized state of the drug-like molecule in octanol, and in water are inputted into the machine learning system.
17. The machine learning based method of claim 1 , in which the machine learning system is used to predict thermodynamic solubility of the drug-like molecule, and in which synthetically generated samples of the thermodynamic ensemble or representation of a solid state of the drug-like molecule and of a dissolved state of the drug-like molecule are inputted into the machine learning system.
18. The machine learning based method of claim 1 , in which the machine learning system is used to predict kinetic solubility of the drug-like molecule, and in which synthetically generated samples of the thermodynamic ensemble or representation of an amorphous solid state of the drug-like molecule and of a dissolved state of the drug-like molecule are inputted into the machine learning system.
19. The machine learning based method of claim 1 , in which the machine learning system is used to predict melting point of the drug-like molecule, and in which synthetically generated samples of the thermodynamic ensemble or representation of a solid state of the drug-like molecule and of the drug-like molecule in octanol are inputted into the machine learning system.
20. The machine learning based method of claim 1 , in which the machine learning system is used to predict acidity (pKa) of the drug-like molecule and in which synthetically generated samples of the thermodynamic ensemble or representation of the drug-like molecule in an appropriate environment are inputted into the machine learning system.
21. The machine learning based method of claim 1 , in which the machine learning system uses Generative adversarial network (GAN) or Variational auto-encoder (VAE), or Graph-Convolutional Policy Network (GCPN) style models.
22. The machine learning based method of claim 1 , in which the machine learning system uses generative models to learn new thermodynamic ensemble or thermodynamic ensemble for which data is not available.
23. The machine learning based method of claim 1 , in which the machine learning system implements weight sharing method when multiple generated samples of thermodynamic ensembles or representations are inputted into the machine learning system.
24. The machine learning based method of claim 1 , in which docking is used to generate the sample of the thermodynamic ensemble or representation.
25. The machine learning based method of claim 1 , in which docking enhanced by molecular dynamics is used to generate the sample of the thermodynamic ensemble or representation.
26. A machine learning based system configured to model a thermodynamic ensemble or representation of a drug-like molecule, in which the machine learning based system is configured to receive and process a synthetically generated sample of the thermodynamic ensemble or representation, the thermodynamic ensemble or representation being a molecular orbital representation or a quantum graph representation of the drug-like molecule; and
in which the quantum graph representation is a molecular graph representation in which each node corresponds to a molecular orbital and edges correspond to a type of quantum correlation between molecular orbitals.
27. A molecule or class of drug-like molecules identified using a machine learning based method of modelling a thermodynamic ensemble or representation of a drug-like molecule, in which a sample of the thermodynamic ensemble or representation is synthetically generated and inputted into a machine learning system, the thermodynamic ensemble or representation being a molecular orbital representation or a quantum graph representation of the drug-like molecule; and
in which the quantum graph representation is a molecular graph representation in which each node corresponds to a molecular orbital and edges correspond to a type of quantum correlation between molecular orbitals.