IP Library › Granted Patent US 12,362,036
Granted Patent B2
US 12,362,036 · App. 16/701,070 · Granted Jul 15, 2025

Protein structure prediction using geometric attention neural networks

Inventors: John Jumper (London, GB); Andrew W. Senior (London, GB); Richard Andrew Evans (London, GB); Stephan Gouws (London, GB); Alexander Bridgland (London, GB)
Assignee: DeepMind Technologies Limited
G16B15/20G16C20/30G16C20/60G16C20/70G16C20/90
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 12,362,036
App. No.
16/701,070
Granted
Jul 15, 2025
Kind
B2
Abstract

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for determining a predicted structure of a protein that is specified by an amino acid sequence. In one aspect, a method comprises: obtaining an initial embedding and initial values of structure parameters for each amino acid in the amino acid sequence, wherein the structure parameters for each amino acid comprise location parameters that specify a predicted three-dimensional spatial location of the amino acid in the structure of the protein; and processing a network input comprising the initial embedding and the initial values of the structure parameters for each amino acid in the amino acid sequence using a folding neural network to generate a network output comprising final values of the structure parameters for each amino acid in the amino acid sequence.

Claims (72)

1. A method performed by one or more computers, the method comprising:

obtaining an initial embedding and initial values of structure parameters for each amino acid in an amino acid sequence of a protein, wherein the structure parameters for each amino acid comprise location parameters that specify a predicted three-dimensional (3D) spatial location of the amino acid in the structure of the protein;

processing a network input comprising the initial embedding and the initial values of the structure parameters for each amino acid in the amino acid sequence using a folding neural network to generate a network output comprising final values of the structure parameters for each amino acid in the amino acid sequence,

wherein the folding neural network comprises a plurality of update blocks, wherein each update block comprises a plurality of neural network layers and is configured to:

receive an update block input comprising a current embedding and current values of the structure parameters for each amino acid in the amino acid sequence; and

process the update block input to update the current embedding and the current values of the structure parameters for each amino acid in the amino acid sequence, comprising:

updating the current embeddings for the amino acids in the amino acid sequence, comprising applying an attention operation over the current embeddings of the amino acids in the amino acid sequence,

wherein the attention operation over the current embeddings of the amino acids in the amino acid sequence: (i) depends at least in part on a 3D geometry defined by the current values of the structure parameters for the amino acids in the amino acids sequence, and (ii) is invariant to rotations and translations over the 3D geometry defined by the current values of the structure parameters for the amino acids in the amino acid sequence;

wherein the final values of the structure parameters for each amino acid in the amino acid sequence collectively characterize a predicted structure of the protein; and

presenting a 3D visual representation of the predicted structure of the protein on a display of a user device.

2. The method of claim 1 , wherein the structure parameters for each amino acid further comprise rotation parameters that specify a predicted spatial orientation of the amino acid in the structure of the protein.

3. The method of claim 2 , wherein the rotation parameters define a 3×3 rotation matrix.

4. The method of claim 1 , wherein processing the update block input to update the current embedding and the current values of the structure parameters for each amino acid in the amino acid sequence further comprises:

updating the current values of the structure parameters for each amino acid in the amino acid sequence based on the updated embeddings for the amino acids in the amino acid sequence.

5. The method of claim 1 , wherein applying the attention operation over the current embeddings of the amino acids in the amino acid sequence comprises:

determining an attention weight for each amino acid in the amino acid sequence, comprising:

generating a three-dimensional query embedding of the given amino acid based on the current embedding of the given amino acid and the current values of the structure parameters for the given amino acid;

generating, for each amino acid in the amino acid sequence, a three-dimensional key embedding of the amino acid based on the current embedding of the amino acid and the current values of the structure parameters for the amino acid; and

determining the attention weight for each amino acid in the amino acid sequence based at least in part on a difference between: (i) the three-dimensional key embedding of the amino acid, and (ii) the three-dimensional query embedding of the given amino acid; and updating the current embedding of the given amino acid using the attention weights.

6. The method of claim 5 , wherein generating the three-dimensional query embedding of the given amino acid based on the current embedding of the given amino acid and the current values of the structure parameters for the given amino acid comprises:

processing the current embedding of the given amino acid using a linear neural network layer that generates a three-dimensional output; and

applying a rotation operation and a translation operation to the three-dimensional output of the linear neural network layer, wherein the rotation operation is specified by current values of rotation parameters for the given amino acid and the translation operation is specified by current values of location parameters for the given amino acid.

7. The method of claim 5 , wherein generating a three-dimensional key embedding of an amino acid based on the current embedding of the amino acid and the current values of the structure parameters for the amino acid comprises:

processing the current embedding of the amino acid using a linear neural network layer that generates a three-dimensional output; and

applying a rotation operation and a translation operation to the three-dimensional output of the linear neural network layer, wherein the rotation operation is specified by current values of rotation parameters for the amino acid and the translation operation is specified by current values of location parameters for the amino acid.

8. The method of claim 5 , wherein updating the current embedding of the given amino acid using the attention weights comprises:

generating, for each amino acid in the amino acid sequence, a three-dimensional value embedding of the amino acid based on the current embedding of the amino acid;

determining a weighted linear combination of the three-dimensional value embeddings of the amino acids using the attention weights;

generating a geometric return embedding by applying a rotation operation and a translation operation to the weighted linear combination, wherein the rotation operation inverts a rotation operation specified by current values of rotation parameters for the given amino acid and the translation operation is specified by a negative of current values of location parameters for the amino acid; and

updating the current embedding of the given amino acid based on the geometric return embedding.

9. The method of claim 5 , wherein updating the current values of the structure parameters for each amino acid in the amino acid sequence based on the updated embeddings for the amino acids in the amino acid sequence comprises, for each amino acid:

determining updated values of location parameters for the amino acid as a sum of: (i) current values of the location parameters for the amino acid, and (ii) a linear projection of the updated embedding of the amino acid;

determining updated values of rotation parameters for the amino acid as a composition of: (i) a rotation operation specified by current values of the rotation parameters for the amino acid, and (ii) a rotation operation specified by a quaternion with real part 1 and imaginary part specified by a linear projection of the updated embedding of the amino acid.

10. The method of claim 5 , further comprising determining the attention weights for the amino acids in the amino acid sequence based data characterizing predicted distances between pairs of amino acids in the structure of the protein.

11. The method of claim 10 , wherein the data characterizing predicted distances between pairs of amino acids in the structure of the protein comprises, for each pair of amino acids, a probability distribution over a set of possible distance ranges between the pair of amino acids.

12. A system comprising:

one or more computers; and

one or more storage devices communicatively coupled to the one or more computers, wherein the one or more storage devices store instructions that, when executed by the one or more computers, cause the one or more computers to perform operations comprising:

obtaining an initial embedding and initial values of structure parameters for each amino acid in an amino acid sequence of a protein, wherein the structure parameters for each amino acid comprise location parameters that specify a predicted three-dimensional (3D) spatial location of the amino acid in the structure of the protein;

processing a network input comprising the initial embedding and the initial values of the structure parameters for each amino acid in the amino acid sequence using a folding neural network to generate a network output comprising final values of the structure parameters for each amino acid in the amino acid sequence,

wherein the folding neural network comprises a plurality of update blocks, wherein each update block comprises a plurality of neural network layers and is configured to:

receive an update block input comprising a current embedding and current values of the structure parameters for each amino acid in the amino acid sequence; and

process the update block input to update the current embedding and the current values of the structure parameters for each amino acid in the amino acid sequence, comprising:

updating the current embeddings for the amino acids in the amino acid sequence, comprising applying an attention operation over the current embeddings of the amino acids in the amino acid sequence,

wherein the attention operation over the current embeddings of the amino acids in the amino acid sequence: (i) depends at least in part on a 3D geometry defined by the current values of the structure parameters for the amino acids in the amino acids sequence, and (ii) is invariant to rotations and translations over the 3D geometry defined by the current values of the structure parameters for the amino acids in the amino acid sequence;

wherein the final values of the structure parameters for each amino acid in the amino acid sequence collectively characterize a predicted structure of the protein; and

presenting a 3D visual representation of the predicted structure of the protein on a display of a user device.

13. The system of claim 12 , wherein the structure parameters for each amino acid further comprise rotation parameters that specify a predicted spatial orientation of the amino acid in the structure of the protein.

14. The system of claim 13 , wherein the rotation parameters define a 3×3 rotation matrix.

15. The system of claim 12 , wherein processing the update block input to update the current embedding and the current values of the structure parameters for each amino acid in the amino acid sequence further comprises:

updating the current values of the structure parameters for each amino acid in the amino acid sequence based on the updated embeddings for the amino acids in the amino acid sequence.

16. The system of claim 12 , wherein applying the attention operation over the current embeddings of the amino acids in the amino acid sequence comprises:

determining an attention weight for each amino acid in the amino acid sequence, comprising:

generating a three-dimensional query embedding of the given amino acid based on the current embedding of the given amino acid and the current values of the structure parameters for the given amino acid;

generating, for each amino acid in the amino acid sequence, a three-dimensional key embedding of the amino acid based on the current embedding of the amino acid and the current values of the structure parameters for the amino acid; and

determining the attention weight for each amino acid in the amino acid sequence based at least in part on a difference between: (i) the three-dimensional key embedding of the amino acid, and (ii) the three-dimensional query embedding of the given amino acid; and

updating the current embedding of the given amino acid using the attention weights.

17. One or more non-transitory computer storage media storing instructions that when executed by one or more computers cause the one or more computers to perform operations comprising:

obtaining an initial embedding and initial values of structure parameters for each amino acid in an amino acid sequence of a protein, wherein the structure parameters for each amino acid comprise location parameters that specify a predicted three-dimensional (3D) spatial location of the amino acid in the structure of the protein;

processing a network input comprising the initial embedding and the initial values of the structure parameters for each amino acid in the amino acid sequence using a folding neural network to generate a network output comprising final values of the structure parameters for each amino acid in the amino acid sequence,

wherein the folding neural network comprises a plurality of update blocks, wherein each update block comprises a plurality of neural network layers and is configured to:

receive an update block input comprising a current embedding and current values of the structure parameters for each amino acid in the amino acid sequence; and

process the update block input to update the current embedding and the current values of the structure parameters for each amino acid in the amino acid sequence, comprising:

updating the current embeddings for the amino acids in the amino acid sequence, comprising applying an attention operation over the current embeddings of the amino acids in the amino acid sequence,

wherein the attention operation over the current embeddings of the amino acids in the amino acid sequence: (i) depends at least in part on a 3D geometry defined by the current values of the structure parameters for the amino acids in the amino acids sequence, and (ii) is invariant to rotations and translations over the 3D geometry defined by the current values of the structure parameters for the amino acids in the amino acid sequence;

wherein the final values of the structure parameters for each amino acid in the amino acid sequence collectively characterize a predicted structure of the protein; and

presenting a 3D visual representation of the predicted structure of the protein on a display of a user device.

18. The non-transitory computer storage media of claim 17 , wherein the structure parameters for each amino acid further comprise rotation parameters that specify a predicted spatial orientation of the amino acid in the structure of the protein.

19. The non-transitory computer storage media of claim 18 , wherein the rotation parameters define a 3×3 rotation matrix.

20. The non-transitory computer storage media of claim 17 , wherein processing the update block input to update the current embedding and the current values of the structure parameters for each amino acid in the amino acid sequence comprises:

wherein processing the update block input to update the current embedding and the current values of the structure parameters for each amino acid in the amino acid sequence further comprises:

updating the current values of the structure parameters for each amino acid in the amino acid sequence based on the updated embeddings for the amino acids in the amino acid sequence.

Assignments (2)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Jun 6, 2025
From: DEEPMIND TECHNOLOGIES LIMITED
To: GDM HOLDING LLC
Reel/Frame 071498/0210 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Mar 6, 2020
From: JUMPER, JOHN; SENIOR, ANDREW W.; EVANS, RICHARD ANDREW; GOUWS, STEPHAN; BRIDGLAND, ALEXANDER
To: DEEPMIND TECHNOLOGIES LIMITED
Reel/Frame 052041/0183 →
Continuity (2)
Provisional Application 62774071 · Nov 30, 2018
Related Publication 20210398606A1 · Dec 23, 2021
References Cited (106)
US 5958784A · Benner · 1999 [cited by applicant]
US 20080215301A1 · Eyal et al. · 2008 [cited by applicant]
US 20130303383A1 · Sander et al. · 2013 [cited by applicant]
US 20130304432A1 · Sander et al. · 2013 [cited by applicant]
US 20170249420A1 · Saladi et al. · 2017 [cited by applicant]
US 20170316147A1 · Gao et al. · 2017 [cited by applicant]
CN 101294970 · 2008 [cited by applicant]
CN 101647022 · 2010 [cited by applicant]
CN 101794351 · 2010 [cited by applicant]
CN 105468934 · 2016 [cited by applicant]
CN 105740646 · 2016 [cited by applicant]
CN 106503484 · 2017 [cited by applicant]
CN 107330512 · 2017 [cited by applicant]
CN 107506613 · 2017 [cited by applicant]
CN 107622182 · 2018 [cited by applicant]
EP 1482433 · 2004 [cited by applicant]
JP 2004258814 · 2004 [cited by applicant]
JP 2022130054 · 2022 [cited by applicant]
WO WO0214875 · 2002 [cited by applicant]
WO WO2017062382 · 2017 [cited by applicant]
Dorn, M. Three-dimensional protein structure prediction: methods and computational strategies. Computational Biology and Chemistry 53: 251-276. (Year: 2014). [cited by examiner]
Du, M. Empirical study of deep neural network architectures for protein secondary structure prediction. Master's Thesis, University of Missouri, 44 pgs. (Year: 2007). [cited by examiner]
Diamond, R. On the multiple simultaneous superposition of molecular structures by rigid body transformations. Protein Science 1: 1279-1287. (Year: 1992). [cited by examiner]
Benros, C. Accessing a novel approach for predicting local 3D protein structures from sequence. Proteins: Structure, Function, and Bioinformatics 62(4): 865-880. (Year: 2006). [cited by examiner]
Hanson AJ. Quaternion maps of global protein structure. Journal of Molecular Graphics and Modelling 38: 256-278. (Year: 2012). [cited by examiner]
Drori I. High quality prediction of protein Q8 secondary structure by diverse neural network architectures. arXiv 1811.07143. 10 pgs. (Year: 2018). [cited by examiner]
Pollastri G. Improving the prediction of protein secondary structure in three and eight classes using recurrent neural networks and profiles. Proteins: Structure, Function, and Genetics 47: 228-235. (Year: 2002). [cited by examiner]
Bohr et al., “A novel approach to prediction of the 3-dimensional structures of protein backbones by neural networks,” FEBS Letters, Feb. 1990, 261(1):43-46. [cited by applicant]
Bohr J et al., “Protein Structures from Distance Inequalities,” Journal of Molecular Biology, Jun. 1993, 231(3):861-869. [cited by applicant]
Boomsma et al., “A generative, probabilistic model of local protein structure,” PNAS, Jan. 2008, 105(26):8932-8937. [cited by applicant]
Derevyanko et al., “Deep convolutional networks for quality assessment of protein folds,” Bioinformatics, Jan. 2018, 34(23):4046-4053. [cited by applicant]
Fan et al., “Virtual Ligand Screening Against Comparative Protein Structure Models,” Methods in Molecular Biology, Nov. 2012, 819:105-126. [cited by applicant]
Feig et al., “Computational protein structure refinement: almost there, yet still so far to go: Computational protein structure refinement,” WIREs Computational Molecular Science, Mar. 2017, 7(3):1-24. [cited by applicant]
Fredholm et al., “A Novel Approach to Prediction of the 3-Dimensional Structures of Protein Backbones by Neural Networks,” Advances in Neural Information Processing Systems 3, Dec. 1990, pp. 523-529. [cited by applicant]
Kukic et al., “Toward an accurate prediction of inter residue distances in proteins using 2D recursive neural networks,” BMC Bioinformatics, Jan. 2014, 15(6): 15 pages. [cited by applicant]
Office Action in Canadian Appln. No. 3,110,200, dated Feb. 9, 2022, 5 pages. [cited by applicant]
Office Action in Canadian Appln. No. 3,110,242, dated Feb. 9, 2022, 5 pages. [cited by applicant]
Office Action in Canadian Appln. No. 3,110,395, dated Feb. 9, 2022, 5 pages. [cited by applicant]
Office Action in Indian Appln. No. 202127003862, dated Jan. 7, 2022, 9 pages. [cited by applicant]
Reczko et al., “Recurrent Neural Networks for Protein Distance Matrix Prediction,” IOS Press, Dec. 1994, pp. 88-97. [cited by applicant]
Sali et al., “Comparative Protein Modelling by Satisfaction of Spatial Restraints,” J. Mol. Biol., Dec. 1993, 234(3):779-815. [cited by applicant]
Uziela et al., “ProQ3D: improved model quality assessments using deep learning,” Bioinformatics, Jan. 2017, 33(10):1578-1580. [cited by applicant]
Wang et al., “Accurate De Novo Prediction of Protein Contact Map by Ultra-Deep Learning Model,” PLOS Computational Biology, Jan. 2017, 13(1):e1005324. [cited by applicant]
Zhao et al., “A Position-Specific Distance-Dependent Statistical Potential for Protein Structure and Functional Study,” Structure, Apr. 2012, 20(6):1118-1126. [cited by applicant]
Zhu et al., “Protein threading using residue co-variation and deep learning,” Bioinformatics, Jun. 2018, 34(13):1263-1273. [cited by applicant]
Defay et al., “Evaluation of Current Techniques for Ab Initio Protein Structure Prediction,” Proteins: Structure, Function, and Genetics, 1995, 23:431-445. [cited by applicant]
Knighton et al., “Crystal Structure of the Catalytic Subunit of Cyclic Adenosine Monophosphate-Dependent Protein Kinase,” Science, Jul. 1991, 253:5018:407-414. [cited by applicant]
Prodromou et al., “Identification and Structural Characterization of the ATP/ADP-Binding Site in the Hsp90 Molecular Chaperone,” Cell, Jul. 1991, 90:65-75. [cited by applicant]
Thornton et al., “Prediction of Progress at Least,” Nature, Nov. 1991, 354:105-106. [cited by applicant]
AlQuraishi, “End-to-end differentiable learning of protein structure,” bioRxiv, Feb. 2018, 36 pages. [cited by applicant]
Ba et al., “Layer Normalization,” https://arxiv.org/abs/1607.06450, Jul. 2016, 14 pages. [cited by applicant]
Bahdanau et al., “Neural machine translation by jointly learning to align and translate,” https://arxiv.org/abs/1409.0473v4, Dec. 2014, 15 pages. [cited by applicant]
Bengio et al., “Can active memory replace attention?” In Advances in Neural Information Processing Systems, (NIPS), Dec. 2016, 9 pages, First author is Kaiser. [cited by applicant]
Britz et al., “Massive exploration of neural machine translation architectures,” https://arxiv.org/abs/1703.03906v2, Mar. 2017, 9 pages. [cited by applicant]
Cheng et al., “Long short-term memory-networks for machine reading,” arXiv preprint arXiv:1601.06733, Sep. 2016, 11 pages. [cited by applicant]
Cho et al., “Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation,” https://arxiv.org/abs/1406.1078v3, last revised Sep. 2014, 15 pages. [cited by applicant]
Chollet, “Xception: Deep Learning with Depthwise Separable Convolutions,” https://arxiv.org/abs/1610.02357v2, Oct. 2016, 14 pages. [cited by applicant]
Chung et al., “Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling,” https://arxiv.org/abs/1412.3555, Dec. 2014, 9 pages. [cited by applicant]
Gehring et al., “Convolutional sequence to sequence learning,” arXiv preprint arXiv:1705.03122v2, Jul. 2017, 15 pages. [cited by applicant]
Graves, “Generating Sequences With Recurrent Neural Networks,” https://arxiv.org/abs/1308.0850v2, Aug. 2013, 43 pages. [cited by applicant]
He et al., “Deep Residual Learning for Image Recognition,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, 770-778. [cited by applicant]
Hochreiter et al., “Gradient flow in recurrent nets: the difficulty of learning long-term dependencies,” bioinf.jku.at, 2001, 15 pages. [cited by applicant]
Hochreiter et al., “Long short-term memory,” Neural Computation, Nov. 1997, 9(8):1735-1780, 32 pages. [cited by applicant]
Ingraham et al., “Learning protein structure with a differentiable simulator,” International Conference on Learning Representations, Feb. 2019, 24 pages. [cited by applicant]
Jozefowicz et al., “Exploring the Limits of Language Modeling,” https://arxiv.org/abs/1602.02410, Feb. 2016, 11 pages. [cited by applicant]
Kaiser et al., “Neural GPUs learn algorithms,” In International Conference on Learning Representations (ICLR), Mar. 2016, 9 pages. [cited by applicant]
Kalchbrenner et al., “Neural machine translation in linear time,” arXiv preprint arXiv:1610.10099v2, Mar. 2017, 9 pages. [cited by applicant]
Kim et al., “Structured attention networks,” https://arxiv.org/abs/1702.00887, Feb. 2017, 21 pages. [cited by applicant]
Kingma et al., “Adam: A Method for Stochastic Optimization,” https://arxiv.org/abs/1412.6980v8, Jul. 2015, 15 pages. [cited by applicant]
Kuchaiev et al., “Factorization tricks for LSTM networks,” https://arxiv.org/abs/1703.10722v2, May 2017, 5 pages. [cited by applicant]
Lin et al., “A Structured Self-attentive Sentence Embedding,” https://arxiv.org/abs/1703.03130, Mar. 2017. [cited by applicant]
Luong et al., “Effective approaches to attention based neural machine translation,” arXiv preprint arXiv:1508.04025, Sep. 2015, 11 pages. [cited by applicant]
Mariani et al., “IDDT: a local superposition-free score for comparing protein structures and models using distance difference tests,” Bioinformatics, Aug. 2013, 29(21):2722-2728. [cited by applicant]
Moré et al., “Distance geometry optimization for protein structures,” Journal of Global Optimization, Aug. 1999, 15(3):219-234. [cited by applicant]
Parikh et al., “A Decomposable Attention Model for Natural Language Inference,” Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, Nov. 2016, 2249-2255. [cited by applicant]
Paulus et al., “A Deep Reinforced Model for Abstractive Summarization,” https://arxiv.org/abs/1705.04304, last revised 2017, 12 page. [cited by applicant]
Press et al., “Using the Output Embedding to Improve Language Models,” https://arxiv.org/abs/1608.05859v2, Nov. 2016, 11 pages. [cited by applicant]
Sennrich et al., “Neural Machine Translation of Rare Words with Subword Units,” https://arxiv.org/abs/1508.07909v1, Aug. 2015, 11 pages. [cited by applicant]
Shazeer et al., “Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer,” https://arxiv.org/abs/1701.06538, Jan. 2017, 19 pages. [cited by applicant]
Srivastava et al., “Dropout: a simple way to prevent neural networks from overfitting,” Journal of Machine Learning Research, Jun. 2014, 15(1):1929-1958. [cited by applicant]
Sukhbaatar et al., “End-to-end memory networks,” Advances in Neural Information Processing Systems, Jan. 2015, 9 pages. [cited by applicant]
Sutskever et al., “Sequence to sequence learning with neural networks,” In Advances in Neural Information Processing Systems, pp. 3104-3112, 2014, 9 pages. [cited by applicant]
Szegedy et al., “Rethinking the inception architecture for computer vision,” https://arxiv.org/abs/1512.00567, Dec. 2015, 10 pages. [cited by applicant]
Vaswani et al., “Attention Is All You Need,” 31st Conference on Neural Information Processing Systems (NIPS 2017), Jun. 2017, 11 pages. [cited by applicant]
Wikipedia, “Quaternions and spatial rotation,” Wikipedia, retrieved from the internet on Dec. 10, 2019 at URL <https://en.wikipedia.org/wiki/Quaternions_and_spatial_rotation>, 13 pages. [cited by applicant]
Wu et al., “Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation,” https://arxiv.org/abs/1609.08144, Oct. 2016, 23 pages. [cited by applicant]
Zhou et al., “Deep recurrent models with fast-forward connections for neural machine translation,” Transactions of the Association for Computational Linguistics, Jul. 2016, 4: 371-383. [cited by applicant]
PCT International Preliminary Report on Patentability in Appln. No. PCT/EP2019/074670, dated Mar. 23, 2021, 17 pages. [cited by applicant]
PCT International Preliminary Report on Patentability in Appln. No. PCT/EP2019/074674, dated Mar. 23, 2021, 16 pages. [cited by applicant]
PCT International Preliminary Report on Patentability in Appln. No. PCT/EP2019/074676, dated Mar. 23, 2021, 16 pages. [cited by applicant]
Decision to Grant Patent in Japanese Appln. No. 2022-130054, dated Oct. 23, 2023, 5 pages (with English translation). [cited by applicant]
Office Action in Japanese Appln. No. 2022-130054, dated Jul. 10, 2023, 6 pages (with English translation). [cited by applicant]
Sadanandan et al., “Automated training of deep convolutional neural networks for cell segmentation,” Scientific reports, 2017, 7.1:7860. [cited by applicant]
Xue et al., “Real-value prediction of backbone torsion angles,” Proteins: Structure, Function, and Bioinformatics, 2008, 72.1:427-433. [cited by applicant]
Notice of Allowance in Chinese Appln. No. 201980054190.6, dated May 15, 2024, 6 pages (with English translation). [cited by applicant]
Office Action in Chinese Appln. No. 201980054171.3, dated May 17, 2024, 14 pages (with English translation). [cited by applicant]
Extended Search Report in European Appln. No. 24180923.5, dated Sep. 12, 2024, 17 pages. [cited by applicant]
Ferreira, Leonardo G., et al. “Molecular docking and structure-based drug design strategies.” Molecules 20.7 (2015): 13384-13421. (Year: 2015). [cited by applicant]
Qian, N., 1999. On the momentum term in gradient descent learning algorithms. Neural networks, 12(1), pp. 145-151. (Year: 1999). [cited by applicant]
Rana, P. S., Sharma, H., Bhattacharya, M. and Shukla, A., 2015. Quality assessment of modeled protein structure using physicochemical properties. Journal of bioinformatics and computational biology, 13(02): 1550005, pp.… [cited by applicant]
Al-Lazikani et al., “Protein structure prediction,” Current Opinion in Chemical Biology, 2001, 5:51-56. [cited by applicant]
Deng et al., “Protein Structure Prediction,” Int. J. Mod. Phys. B, Jul. 2018, 32(18):18 pages. [cited by applicant]
Office Action in Chinese Appln. No. 201980054143.1, dated Dec. 19, 2023, 29 pages (with English translation). [cited by applicant]
Office Action in Chinese Appln. No. 201980054171.3, dated Dec. 28, 2023, 34 pages (with English translation). [cited by applicant]
Wang et al., “New Deep Neural Networks for Protein Model Evaluation,” 2017 International Conference on Tools with Artificial Intelligence, Nov. 6, 2017, pp. 309-313. [cited by applicant]
Gao, Y., Wang, S., Deng, M. and Xu, J., Jan. 17, 2018. RaptorX-Angle: real-value prediction of protein backbone dihedral angles through a hybrid method of clustering and deep learning. BMC bioinformatics, 19, pp. 73-84.… [cited by applicant]