IP Library › Granted Patent US 12,694,953
Granted Patent B2
US 12,694,953 · App. 17/427,786 · Granted Jul 28, 2026

Systems, methods, and media for molecule design using machine learning mechanisms

Inventors: Neal W. Woodbury (Tempe, AZ); Alexander T. Taguchi (Cambridge, MA)
Assignee: ARIZONA BOARD OF REGENTS ON BEHALF OF ARIZONA STATE UNIVERSITY
G16C20/50G16B40/10G16B40/20
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,694,953
App. No.
17/427,786
Filed
Aug 2, 2021
Granted
Jul 28, 2026
Kind
B2
Art Unit
1635
USPC
703/12
Abstract

Mechanisms for molecule design using machine learning include: forming a first training set for a neural network using, for each of a first plurality of known molecules, a plurality of input values that represent the structure of the known molecule and a plurality of functional property values for the known molecule; training the neural network using the first training set; proposing a first plurality of proposed molecules, and predicting first predicted functional property values of the first plurality of proposed molecules that have the desired function property values; causing the first plurality of proposed molecules to be synthesized to form a first plurality of synthesized molecules; receiving first measured functional property values of the first plurality of synthesized molecules; and adding data regarding the first plurality of synthesized molecules to the first training set to form a second training set and retrain the neural network using the second training set.

Claims (82)

1 . A system for proposing molecules having desired functional property values, comprising:

a computerized synthesizer;

a memory; and

a hardware processor coupled to the computerized synthesizer and to the memory and configured to:

form a first training set for a neural network by:

selecting features of building block molecules to produce selected features and functional properties of a first plurality of proposed molecules that comprise the building block molecules to produce selected functional properties;

accessing a library that identifies known molecules A known made from the building block molecules;

forming a representation of the library in which each known molecule A known is shown as a matrix B known of representations of the building block molecules;

forming a matrix T of the values V of the selected features of all of the building block molecules in the library or all of features of the building block molecules;

multiplying each matrix B known by the matrix T to form a matrix C known of features;

linearizing each matrix C known to form vectors D known ; and

determining, for each vector D known , known values P known for corresponding known molecules A known that comprise the selected functional properties;

train the neural network using at least the vectors D known of the first training set to predict values P known of the selected function properties to produce a trained neural network;

propose a first plurality of proposed molecules A proposed , and predict values P proposed of the selected functional properties of the proposed molecules A proposed using the trained neural network;

synthesize selected molecules of the first plurality of proposed molecules A proposed to form a first plurality of synthesized molecules A synth using the computerized synthesizer;

receive first measured functional property values of the first plurality of synthesized molecules A synth ; and

add data regarding selected molecules of the first plurality of synthesized molecules A synth to the library used to form the first training set to form a second training set and retrain the neural network using the second training set.

2 . The system of claim 1 , wherein the representation of the library is based on one-hot representations of the building block molecules that form the known molecules A known , wherein the one-hot representations are formed using one-hot encoding of the building block molecules.

3 . The system of claim 2 , wherein the building block molecules are amino acids.

4 . The system of claim 1 , wherein the known molecules A known are peptides.

5 . The system of claim 1 , wherein the representation of the library is based on chemical properties of building block molecules that form the known molecules A known .

6 . The system of claim 1 , wherein the neural network includes an encoder layer based on chemical properties of building block molecules that form the known molecules A known .

7 . The system of claim 1 , wherein an iterative process is used to propose the first plurality of proposed molecules.

8 . The system of claim 7 , wherein the iterative process attempts to find a local maximum based on each of the first plurality of proposed molecules.

9 . The system of claim 1 , wherein the hardware processor is further configured to:

propose a second plurality of proposed molecules, and predict second predicted functional property values of the second plurality of proposed molecules that have desired function property values;

synthesize selected molecules of the second plurality of proposed molecules to form a second plurality of synthesized molecules using the computerized synthesizer;

receive second measured functional property values of the second plurality of synthesized molecules; and

determine whether the second measured functional property values a threshold amount different from a first measured functional property values.

10 . A method for proposing molecules having desired functional property values, comprising:

forming a first training set for a neural network by:

selecting features of building block molecules to produce selected features and functional properties of a first plurality of proposed molecules that comprise the building block molecules to produce selected functional properties;

accessing a library that identifies known molecules A known made from the building block molecules;

forming a representation of the library in which each known molecule A known is shown as a matrix B known of representations of the building block molecules;

forming a matrix T of the values V of the selected features of all of the building block molecules in the library or all of features of the building block molecules;

multiplying each matrix B known by the matrix T to form a matrix C known of features;

linearizing each matrix C known to form vectors D known ; and

determining, for each vector D known , known values P known for corresponding known molecules A known that comprise the selected functional properties;

training the neural network using at least the vectors D known of the first training set using a hardware processor to predict values P known of the selected function properties to produce a trained neural network;

proposing a first plurality of proposed molecules A proposed , and predict values P proposed of the selected functional properties of the proposed molecules A proposed using the trained neural network;

synthesizing selected molecules of the first plurality of proposed molecules A proposed to form a first plurality of synthesized molecules A synth using a computerized synthesizer;

receiving first measured functional property values of the first plurality of synthesized molecules A synth ; and

adding data regarding selected molecules of the first plurality of synthesized molecules A synth to the library used to form the first training set to form a second training set and retrain the neural network using the second training set.

11 . The method of claim 10 , wherein the representation of the library is based on one-hot representations of the building block molecules that form the known molecules A known , wherein the one-hot representations are formed using one-hot encoding of the building block molecules.

12 . The method of claim 11 , wherein the building block molecules are amino acids.

13 . The method of claim 10 , wherein the known molecules A known are peptides.

14 . The method of claim 10 , wherein the representation of the library is based on chemical properties of building block molecules that form the known molecules A known .

15 . The method of claim 10 , wherein the neural network includes an encoder layer based on chemical properties of building block molecules that form the known molecules A known .

16 . The method of claim 10 , wherein an iterative process is used to propose the first plurality of proposed molecules.

17 . The method of claim 16 , wherein the iterative process attempts to find a local maximum based on each of the first plurality of proposed molecules.

18 . The method of claim 10 , further comprising:

propose a second plurality of proposed molecules, and predict second predicted functional property values of the second plurality of proposed molecules that have desired function property values;

synthesize selected molecules of the second plurality of proposed molecules to form a second plurality of synthesized molecules using the computerized synthesizer;

receive second measured functional property values of the second plurality of synthesized molecules; and

determine whether the second measured functional property values a threshold amount different from a first measured functional property values.

19 . A non-transitory computer-readable medium containing computer executable instructions that, when executed by a processor, cause the processor to perform a method proposing molecules having desired functional property values, the method comprising:

forming a first training set for a neural network by:

selecting features of building block molecules to produce selected features and functional properties of a first plurality of proposed molecules that comprise the building block molecules to produce selected functional properties;

accessing a library that identifies known molecules A known made from the building block molecules;

forming a representation of the library in which each known molecule A known is shown as a matrix B known of representations of the building block molecules;

forming a matrix T of the values V of the selected features of all of the building block molecules in the library or all of features of the building block molecules;

multiplying each matrix B known by the matrix T to form a matrix C known of features;

linearizing each matrix C known to form vectors D known ; and

determining, for each vector D known , known values P known for corresponding known molecules A known that comprise the selected functional properties;

training the neural network using at least the vectors D known of the first training set using a hardware processor to predict values P known of the selected function properties to produce a trained neural network;

proposing a first plurality of proposed molecules A proposed , and predict values P proposed of the selected functional properties of the proposed molecules A proposed using the trained neural network;

synthesizing selected molecules of the first plurality of proposed molecules A proposed to form a first plurality of synthesized molecules A synth using a computerized synthesizer;

receiving first measured functional property values of the first plurality of synthesized molecules A synth ; and

adding data regarding selected molecules of the first plurality of synthesized molecules A synth to the library used to form the first training set to form a second training set and retrain the neural network using the second training set.

20 . The non-transitory computer-readable medium of claim 19 ,

wherein the representation of the library is based on one-hot representations of the building block molecules that form the known molecules A known , wherein the one-hot representations are formed using one-hot encoding of the building block molecules.

21 . The non-transitory computer-readable medium of claim 20 , wherein the building block molecules are amino acids.

22 . The non-transitory computer-readable medium of claim 19 , wherein the known molecules A known are peptides.

23 . The non-transitory computer-readable medium of claim 19 , wherein the representation of the library is based on chemical properties of building block molecules that form the known molecules A known .

24 . The non-transitory computer-readable medium of claim 19 , wherein the neural network includes an encoder layer based on chemical properties of building block molecules that form the known molecules A known .

25 . The non-transitory computer-readable medium of claim 19 , wherein an iterative process is used to propose the first plurality of proposed molecules.

26 . The non-transitory computer-readable medium of claim 25 , wherein the iterative process attempts to find a local maximum based on each of the first plurality of proposed molecules.

27 . The non-transitory computer-readable medium of claim 19 , wherein the method further comprises:

propose a second plurality of proposed molecules, and predict second predicted functional property values of the second plurality of proposed molecules that have desired function property values;

synthesize selected molecules of the second plurality of proposed molecules to form a second plurality of synthesized molecules using the computerized synthesizer;

receive second measured functional property values of the second plurality of synthesized molecules; and

determine whether the second measured functional property values a threshold amount different from a first measured functional property values.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Aug 14, 2023
From: WOODBURY, NEAL W.; TAGUCHI, ALEXANDER T.
To: ARIZONA BOARD OF REGENTS ON BEHALF OF ARIZONA STATE UNIVERSITY
Reel/Frame 064578/0040 →
Continuity (2)
Provisional Application 62804029 · Feb 11, 2019
Related Publication 20220130494A1 · Apr 28, 2022
References Cited (73)
US 5434796A · Weininger · 1995 [cited by applicant]
US 20060206269A1 · Diller et al. · 2006 [cited by applicant]
US 20120065123A1 · Johnston et al. · 2012 [cited by applicant]
US 20150278441A1 · Min et al. · 2015 [cited by applicant]
US 20160176951A1 · Barelle et al. · 2016 [cited by applicant]
US 20170161635A1 · Oono et al. · 2017 [cited by applicant]
Aditya et al., DeepIU: An Architecture for Image Understanding, Advances in Cognitive Systems 4, 2016, 16 pages. [cited by applicant]
Aditya et al., Answering Image Riddles using Vision and Reasoning through Probabilistic Soft Logic, arXiv:1611.05896v1 [cs.CV], Nov. 17, 2016, 14 pages. [cited by applicant]
Aditya et al., Explicit Reasoning over End-to-End Neural Architectures for Visual Question Answering, The Thirty-Second AAAI Conference on Artificial Intelligence (AAAI-18), pp. 629-637. [cited by applicant]
Alikhani et al., Arrows are the Verbs of Diagrams, Proceeding of the 27th International Conference on Computational Linguistics, Aug. 20-26, 2018, pp. 3552-3563. [cited by applicant]
Alquraishi, End-to-End Differentiable Learning of Protein Structure, Cell Systems 8, Apr. 24, 2019, pp. 292-301. [cited by applicant]
Amanchy et al., Stable Isotope Labeling with Amino Acids in Cell Culture (SILAC) for Studying Dynamics of Protein Abundance and Posttranslational Modifications, Science's stke, Protocol, Jan. 18, 2005, 20 pages. [cited by applicant]
Belkin et al., Laplacian Eigenmaps for Dimensionality Reduction and Data Representation, Neural Computation, 15, 2003, pp. 1373-1396. [cited by applicant]
Berisha et al., Empirically Estimable Classification Bounds Based on a Nonparametric Divergence Measure, IEEE Transactions on Signal Processing, vol. 64, No. 3, Feb. 1, 2016, pp. 580-591. [cited by applicant]
Choudhury et al., Automatic Extraction of Figures from Scholarly Documents, DocEng '15, Sep. 8-11, 2015, pp. 47-50. [cited by applicant]
Choudhury et al., An Architecture for Information Extraction from Figures in Digital Libraries, WWWWW 2015 Companion, May 18-22, 2015, pp. 667-672. [cited by applicant]
Clark et al., PDFFigures 2.0: Mining Figures from Research Papers, JCDL '16, Jun. 19-23, 2016 pp. 143-152. [cited by applicant]
Seo et al., Diagram Understanding in Geometry Questions, Proceeding of the Twenty-Eighth AAAI Conference on Artificial Intelligence, 2014, pp. 2831-2838. [cited by applicant]
Dimaio et al., Modeling Symmetric Macromolecular Structures in Rosetta3, PLoS One, vol. 6, Issue 6, Jun. 2011, 13 pages. [cited by applicant]
Evans et al., De novo structure prediction with deep-learning based scoring, ResearchGate, Dec. 2018, 3 pages. [cited by applicant]
Clark et al., Looking Beyond Text: Extracting Figures, Tables, and Captions from Computer Science Papers, Scholary Big Data: AI Perspectives, Challenges, and Ideas: Papers from the 2015 AAAI Workshop, 7 pages. [cited by applicant]
Fout et al., Protein Interface Prediction using Graph Convolutional Networks, 31st Conference on Neural Information Processing Systems (NIPS 2017), 10 pages. [cited by applicant]
Gao et al., Destini: A deep-learning approach to contact-driven protein structure prediction, Scientific Reports, 9:3514, 2019, 13 pages. [cited by applicant]
Gfeller et al., SwissSidechain: a molecular and structural database of non-natural sidechains, Nucleic Acids Research, 2013, vol. 41, Oct. 26, 2012, Database issue pp. D327-D332. [cited by applicant]
Ghodsi, Dimensionality Reduction A Short Tutorial, Department of Statistics and Actuarial Science University of Waterloo, 2006, 25 pages. [cited by applicant]
Grover et al., Flow-GAN: Bridging implicit and prescribed learning in generative models, ResearchGate, May 2017, 7 pages. [cited by applicant]
Guye et al., Genetically engineering self-organization of human pluripotent stem cells into a liver bud-like tissue using Gata6, Nature Communications, 7:10243, 12 pages. [cited by applicant]
Haney et al., Reassessing the Host Defense Peptide Landscape, Frontiers in Chemistry, vol. 7, Article 43, Feb. 2019, 22 pages. [cited by applicant]
Hashemifar et al., Predicting protein-protein interactions through sequence-based deep learning, Bioinformatics, 34, 2018, pp. i802-i810. [cited by applicant]
Hobbs et al., Toward a Formal Theory of Information Structure, USC Information Sciences Institute, 25 pages. [cited by applicant]
Ioffe et al., Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift, Google, 1600 Amphitheatre Pkwy, Mountain View, CA 94043, 9 pages. [cited by applicant]
Jain et al., Low-rank Matrix Completion using Alternating Minimization, STOC '13, Jun. 1-4, 2013, pp. 665-674. [cited by applicant]
Jenson et al., Peptide design by optimization on a data-parameterized protein interaction landscape, PNAS, vol. 115, No. 44, pp. E10342-E10351. [cited by applicant]
Jespersen et al., Antibody Specific B-Cell Epitope Predictions: Leveraging Information From Antibody-Antigen Protein Complexes, Frontiers in Immunology, vol. 10, Article 298, Feb. 2019, 10 pages. [cited by applicant]
Johansson-Akhe, et al., Predicting protein-peptide interaction sites using distant protein complexes as structural templates, Scientific Reports, 9:4267, 2019, 13 pages. [cited by applicant]
Kembhavi et al., A Diagram is Worth a Dozen Images, ECCV, Part IV, LNCS 9908, 2016, pp. 235-251. [cited by applicant]
Konyk et al., Chemical Knowledge for the Semantic Web, ResearchGate, Jun. 2008, 9 pages. [cited by applicant]
Kurata et al., Structure and Semantics of Arrow Diagrams, COSIT, LNCS 3693, 2005, pp. 232-250. [cited by applicant]
Legutki et al., Scalable high-density peptide arrays for comprehensive health monitoring, Nature Communications, 5:4785, 2014, 7 pages. [cited by applicant]
Liang et al., An Index for Characterization of Natural and Non-Natural Amino Acids for Peptidomimetics, PLoS One, vol. 8, Issue 7, e67844, Jul. 2013, 16 pages. [cited by applicant]
Liu et al., Antibody complementarity determining region design using high-capacity machine learning, Bioinformatics, 36(7), 2020, pp. 2126-2133. [cited by applicant]
Lee et al., Machine learning-enabled discovery and design of membrane-active peptides, Bioorg Med Chem., 26(10), Jun. 1, 2018, pp. 2708-2718. [cited by applicant]
Mayne et al., Rapid parameterization of small molecules using the Force Field Toolkit, J Comput Chem., 34(32), Dec. 15, 2013, 28 pages. [cited by applicant]
Miao et al., Gaussian Accelerated Molecular Dynamics: Unconstrained Enhanced Sampling and Free Energy Calculation, Journal of Chemical Theory and Computation, 11, 2015, pp. 3584-3595. [cited by applicant]
Nair et al., Rectified Linear Units Improve Restricted Boltzmann Machines, Proceedings of the 27th International Conference on Machine Learning, Israel, 2010, 8 pages. [cited by applicant]
Rose et al., The new biology: beyond the Modern Synthesis, Biology Direct, 2:30, Nov. 24, 2007, 17 pages. [cited by applicant]
Rowe et al., An ImmunoSignature test distinguishes Trypanosoma cruzi, hepatitis B, hepatitis C and West Nile virus seropositivity among asymptomatic blood donors, PLOS Neglected Tropical Diseases, Sep. 5, 2017, 30 pages. [cited by applicant]
Ruder, An overview of gradient descent optimization algorithms*, arXiv:1609.04747v1 [cs.LG], Sep. 15, 2016, 12 pages. [cited by applicant]
Sandberg, log, 1998, 1 page. [cited by applicant]
Savva et al., ReVision: Automated Classification, Analysis and Redesign of Chart Images, Paper Session: Patterns, UIST'11, Oct. 16-19, 2011, 10 pages. [cited by applicant]
Schirrmeister et al., Training Generative Reversible Networks, Presented at the ICML 2018 workshop on Theoretical Foundations and Applications of Deep Generative Models, Stockholm, Sweden, PMLR 80, 2018, 10 pages. [cited by applicant]
Siegel et al., FigureSeer: Parsing Result-Figures in Research Papers, Allen Institute for Artificial Intelligence, University of Washington, http://allenai.org/plato/figureseer, 16 pages. [cited by applicant]
Singh et al., Humoral Immunity Profiling of Subjects with Myalgic Encephalomyelitis Using a Random Peptide Microarray Differentiates Cases from Controls with High Specificity and Sensitivity, Mol Neurobiol, 55, 2018, pp… [cited by applicant]
Srivastava et al., Dropout: A Simple Way to Prevent Neural Networks from Overfitting, Journal of Machine Learning Research 15, 2014, pp. 1929-1958. [cited by applicant]
Tallorin et al., Discovering de novo peptide substrates for enzymes using machine learning, Nature Communications, 9:5253, 2018, 10 pages. [cited by applicant]
Tominaga, Comparative study of class data analysis with PCA-LDA, SIMCA, PLS, ANNs, and k-NN, Chemometrics and Intelligent Laboratory Systems, 49, 1999, pp. 105-115. [cited by applicant]
Sharma et al., Towards Addressing the Winograd Schema Challenge—Building and Using a Semantic Parser and a Knowledge Hunting Module, Proceedings of the Twenty-Fourth International Joint Conference on Artificial Intellig… [cited by applicant]
Tu et al., Objective Assessment of Pathological Speech Using Distribution Regression, IEEE, ICASSP, 2017, pp. 5050-5054. [cited by applicant]
Udell et al., Generalized Low Rank Models, Foundations and Trends in Machine Learning, vol. 9, No. 1, 2016, pp. 1-118. [cited by applicant]
Valenzuela-Escarcega et al., Large-scale automated machine reading discovers new cancer-driving mechanisms, Database, 2018, pp. 1-14. [cited by applicant]
Vanommeslaeghe et al., Automation of the CHARMM General Force Field (CGenFF) I: Bond Perception and Atom Typing, Journal of Chemical Information and Modeling, 52, 2012, pp. 3144-3154. [cited by applicant]
Velazquez et al., Programming Morphogenesis through Systems and Synthetic Biology, Trends in Biotechnology, vol. 36, No. 4, Apr. 2018, pp. 415-429. [cited by applicant]
Veltri et al., Deep learning improves antimicrobial peptide recognition, Bioinformatics, 34(16), 2018, pp. 2740-2747. [cited by applicant]
Vo et al., The NL2KR Platform for building Natural Language Translation Systems, Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natu… [cited by applicant]
Wang, Closed-form Inverse Kinematic Solution for Anthropomorphic Motion in Redundant Robot Arms, A Thesis Presented in Partial Fulfillment of the Requirements for the Degree Master of Science, Arizona State University, … [cited by applicant]
Warren et al., On the Control of Human-Robot Bi-Manual Manipulation, J Intell Robot Syst, 78, 2015, pp. 21-32. [cited by applicant]
Williams et al., Creating Protein Affinity Reagents by Combining Peptide Ligands on Synthetic DNA Scaffolds, J. Am. Chem. Soc., 131, 2009, pp. 17233-17241. [cited by applicant]
Wisler et al., Empirically-Estimable Multi-Class Classification Bounds, IEEE, ICAASP, 2016, pp. 2594-2598. [cited by applicant]
Wisler et al., A data-driven basis for direct estimation of functionals of distributions, arXiv:1702.06516v1 [cs.IT], Feb. 21, 2017, 13 pages. [cited by applicant]
Wisler et al., Direct Estimation of Density Functionals Using a Polynomial Basis, IEEE Transactions on Signal Processing, vol. 66, No. 3, Feb. 1, 2018, pp. 558-572. [cited by applicant]
Yang et al., Corpus-Guided Sentence Generation of Natural Images, Proceedings of the 2011 Conference on Empirical Methods in Natural Language Processing, Edinburgh, Scotland, UK, Jul. 27-31, 2011,pp. 444-454. [cited by applicant]
Yang et al., A Cognitive System for Understanding Human Manipulation Actions, Advances in Cognitive Systems, 3, 2014, pp. 67-86. [cited by applicant]
Yoshida et al., Using Evolutionary Algorithms and Machine Learning to Explore Sequence Space for the Discovery of Antimicrobial Peptides, Chem 4, Mar. 8, 2018, pp. 533-543. [cited by applicant]