IP Library › Granted Patent US 12,651,432
Granted Patent B1
US 12,651,432 · App. 18/526,742 · Granted Jun 9, 2026

Utilizing machine learning models to synthesize perturbation data to generate perturbation heatmap graphical user interfaces

Inventors: Marta Marie Fay (Salt Lake City, UT); August Orvis Allen (Boulder, CO); Eugene Yin-Chung Ting (Toronto, CA); Lina Maria Nilsson (Salt Lake City, UT); Condie Thomas Swallow, II (West Valley City, UT); Michael Haines (Salt Lake City, UT); Denton Hallar Greenfield (Evansville, IN); Kristin Ann Clark (Lehi, UT); Lovina Roundy (Orem, UT); Michael Joseph Uloth (Dundas, CA); Sara Marjean Moore (Boise, ID); Shweta Deepchand Bhandare (Boulder, CO); Ted Douglas Monchamp (Nashua, NH); Summer Walid Elias (Salt Lake City, UT); Berton Allen Earnshaw (Cedar Hills, UT); Mason Lemoyne Victors (Riverton, UT); Safiye Celik (Sudbury, MA); James Benjamin Taylor (Midlothian, VA); Andrew David Blevins (Salt Lake City, UT); James Douglas Jensen (Farmington, UT); Jacob Carter Cooper (Sandy, UT); Conor Austin Forsman Tillinghast (Salt Lake City, UT); Seyhmus Guler (Salt Lake City, UT); Kyle Rollins Hansen (Kaysville, UT); Sarah Jordan DeVore (Salt Lake City, UT); Tongzhou Shen (Surrey, CA)
Assignee: Recursion Pharmaceuticals, Inc.
G06V10/761G06F3/0482
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 12,651,432
App. No.
18/526,742
Granted
Jun 9, 2026
Kind
B1
Abstract

The present disclosure relates to systems, non-transitory computer-readable media, and methods for embedding perturbation data via a machine learning model and filtering, aligning, and aggregating the embeddings to generate a genome-wide perturbation database for real-time generation of perturbation heatmaps. In particular, in one or more embodiments, the disclosed systems can receive a plurality of perturbation images portraying cells from a plurality of wells corresponding to a plurality of cell perturbations. Further, the systems can generate, utilizing a machine learning model, a plurality of well-level image embeddings from the plurality of perturbation images. Moreover, the systems can align, utilizing an alignment model, the plurality of well-level image embeddings to generate aligned well-level image embeddings. Additionally, the systems can aggregate, according to perturbations of one or more perturbation experiments, the well-level image embeddings to generate perturbation-level image embeddings. Furthermore, the systems can generate perturbation comparisons utilizing the perturbation-level image embeddings.

Claims (48)

1 . A computer-implemented method comprising:

providing, for display via a user interface of a requestor device, a perturbation query element for selection of a plurality of cell perturbations;

in response to user interaction with the perturbation query element, providing, for display via the user interface of the requestor device, a perturbation heatmap portraying similarity measures from a plurality of perturbation embeddings for the plurality of cell perturbations; and

in response to user interaction with a similarity measure portrayed via the perturbation heatmap, providing, for display on the requestor device, a perturbation data element comprising one or more features of the plurality of cell perturbations corresponding to the similarity measure.

2 . The computer-implemented method of claim 1 , wherein providing the perturbation query element comprises providing, for display via the user interface, a first plurality of selectable options for a plurality of gene knockout perturbations and a second plurality of selectable options for a plurality of compound perturbations, and further comprising determining the plurality of cell perturbations for the perturbation heatmap based on one or more user interactions with the first plurality of selectable options or the second plurality of selectable options.

3 . The computer-implemented method of claim 1 , further comprising determining the similarity measure of the similarity measures by:

determining a first perturbation image embedding for a first perturbation of the plurality of cell perturbations;

determining a second perturbation image embedding for a second perturbation; and

comparing the first perturbation image embedding and the second perturbation image embedding to determine the similarity measure.

4 . The computer-implemented method of claim 3 , wherein providing the perturbation heatmap comprises providing, for display via the user interface of the requestor device, the similarity measure in a cell within a row of the perturbation heatmap corresponding to the first perturbation and within a column of the perturbation heatmap corresponding to the second perturbation.

5 . The computer-implemented method of claim 4 , wherein providing the perturbation data element comprises, providing, for display via the user interface of the requestor device based on the user interaction with the similarity measure, a heatmap overlay element portraying a first set of features corresponding to the first perturbation and a second set of features corresponding to the second perturbation.

6 . The computer-implemented method of claim 3 , further comprising:

providing, for display via the user interface of the requestor device, a similarity measure element for selection of a similarity threshold;

determining that the similarity measure between the first perturbation image embedding and the second perturbation image embedding satisfies the similarity threshold; and

based on the similarity measure satisfying the similarity threshold, selecting the second perturbation to include within the perturbation heatmap.

7 . The computer-implemented method of claim 1 , further comprising:

providing, for display via the user interface of the requestor device, at least one of a perturbation statistical significance threshold element for selection of a statistical significance threshold or a compound concentration element for selection of a compound concentration; and

in response to user interaction with the perturbation statistical significance threshold element or the compound concentration element, providing, for display via the user interface of the requestor device, an updated perturbation heatmap portraying similarity measures between a plurality of perturbation embeddings for the plurality of cell perturbations according to the statistical significance threshold or the compound concentration.

8 . The computer-implemented method of claim 1 , further comprising providing, for display with the perturbation heatmap via the user interface of the requestor device, at least one of: a similarity distribution element for displaying similarity distributions, a gene information element for displaying gene information, or an enrichment element for displaying gene set enrichment analysis.

9 . A system comprising:

at least one processor; and

at least one non-transitory computer-readable storage medium storing instructions that, when executed by the at least one processor, cause the system to:

provide, for display via a user interface of a requestor device, a perturbation query element for selection of a plurality of cell perturbations;

in response to user interaction with the perturbation query element, provide, for display via the user interface of the requestor device, a perturbation heatmap portraying similarity measures from a plurality of perturbation embeddings for the plurality of cell perturbations; and

in response to user interaction with a similarity measure portrayed via the perturbation heatmap, provide, for display on the requestor device, a perturbation data element comprising one or more features of the plurality of cell perturbations corresponding to the similarity measure.

10 . The system of claim 9 , wherein the instructions cause the system to provide the perturbation query element by providing, for display via the user interface, a first plurality of selectable options for a plurality of gene knockout perturbations and a second plurality of selectable options for a plurality of compound perturbations, and further comprising determining the plurality of cell perturbations for the perturbation heatmap based on one or more user interactions with the first plurality of selectable options or the second plurality of selectable options.

11 . The system of claim 9 , further comprising instructions that, when executed by the at least one processor, cause the system to determine the similarity measure of the similarity measures by:

determining a first perturbation image embedding for a first perturbation of the plurality of cell perturbations;

determining a second perturbation image embedding for a second perturbation; and

comparing the first perturbation image embedding and the second perturbation image embedding to determine the similarity measure.

12 . The system of claim 11 , wherein the instructions cause the system to provide the perturbation heatmap by providing, for display via the user interface of the requestor device, the similarity measure in a cell within a row of the perturbation heatmap corresponding to the first perturbation and within a column of the perturbation heatmap corresponding to the second perturbation.

13 . The system of claim 12 , wherein the instructions cause the system to provide the perturbation data element by providing, for display via the user interface of the requestor device based on the user interaction with the similarity measure, a heatmap overlay element portraying a first set of features corresponding to the first perturbation and a second set of features corresponding to the second perturbation.

14 . The system of claim 11 , further comprising instructions that, when executed by the at least one processor, cause the system to:

provide, for display via the user interface of the requestor device, a similarity measure element for selection of a similarity threshold;

determining that the similarity measure between the first perturbation image embedding and the second perturbation image embedding satisfies the similarity threshold; and

based on the similarity measure satisfying the similarity threshold, selecting the second perturbation to include within the perturbation heatmap.

15 . A non-transitory computer-readable medium storing instructions that, when executed by at least one processor, cause a computing device to:

provide, for display via a user interface of a requestor device, a perturbation query element for selection of a plurality of cell perturbations;

in response to user interaction with the perturbation query element, provide, for display via the user interface of the requestor device, a perturbation heatmap portraying similarity measures from a plurality of perturbation embeddings for the plurality of cell perturbations; and

in response to user interaction with a similarity measure portrayed via the perturbation heatmap, provide, for display on the requestor device, a perturbation data element comprising one or more features of the plurality of cell perturbations corresponding to the similarity measure.

16 . The non-transitory computer-readable medium of claim 15 , wherein the instructions cause the computing device to provide the perturbation query element by providing, for display via the user interface, a first plurality of selectable options for a plurality of gene knockout perturbations and a second plurality of selectable options for a plurality of compound perturbations, and further comprising determining the plurality of cell perturbations for the perturbation heatmap based on one or more user interactions with the first plurality of selectable options or the second plurality of selectable options.

17 . The non-transitory computer-readable medium of claim 15 , further comprising instructions that, when executed by the at least one processor, cause the computing device to determine the similarity measure of the similarity measures by:

determining a first perturbation image embedding for a first perturbation of the plurality of cell perturbations;

determining a second perturbation image embedding; and

comparing the first perturbation image embedding and the second perturbation image embedding to determine the similarity measure.

18 . The non-transitory computer-readable medium of claim 17 , wherein the instructions cause the computing device to provide the perturbation heatmap by providing, for display via the user interface of the requestor device, the similarity measure in a cell within a row of the perturbation heatmap corresponding to the first perturbation and within a column of the perturbation heatmap corresponding to the second perturbation.

19 . The non-transitory computer-readable medium of claim 18 , wherein the instructions cause the computing device to provide the perturbation data element by providing, for display via the user interface of the requestor device based on the user interaction with the similarity measure, a heatmap overlay element portraying a first set of features corresponding to the first perturbation and a second set of features corresponding to the second perturbation.

20 . The non-transitory computer-readable medium of claim 15 , further comprising instructions that, when executed by the at least one processor, cause the computing device to provide, for display with the perturbation heatmap via the user interface of the requestor device, at least one of: a similarity distribution element for displaying similarity distributions, a gene information element for displaying gene information, or an enrichment element for displaying gene set enrichment analysis.

Assignments (2)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Dec 1, 2023
From: FAY, MARTA MARIE; ALLEN, AUGUST ORVIS; TING, EUGENE YIN-CHUNG; NILSSON, LINA MARIA; SWALLOW, CONDIE THOMAS, II; HAINES, MICHAEL; GREENFIELD, DENTON HALLAR; CLARK, KRISTIN ANN; ULOTH, MICHAEL JOSEPH; MOORE, SARA MARJEAN; BHANDARE, SHWETA DEEPCHAND; MONCHAMP, TED DOUGLAS; ELIAS, SUMMER WALID; EARNSHAW, BERTON ALLEN; VICTORS, MASON LEMOYNE; CELIK, SAFIYE; TAYLOR, JAMES BENJAMIN; BLEVINS, ANDREW DAVID; JENSEN, JAMES DOUGLAS; COOPER, JACOB CARTER; TILLINGHAST, CONOR AUSTIN FORSMAN; GULER, SEYHMUS; HANSEN, KYLE ROLLINS; DEVORE, SARAH JORDAN; SHEN, TONGZHOU
To: RECURSION PHARMACEUTICALS, INC.
Reel/Frame 065738/0321 →
AGREEMENT Recorded Dec 1, 2023
From: ROUNDY, LOVINA
To: RECURSION PHARMACEUTICALS, INC.
Reel/Frame 065744/0731 →
Continuity (1)
Provisional Application 63582702 · Sep 14, 2023
References Cited (54)
US 7047410B1 · Shin · 2006 [cited by applicant]
US 8812526B2 · Ramer et al. · 2014 [cited by applicant]
US 10146914B1 · Victors et al. · 2018 [cited by applicant]
US 10769501B1 · Ando · 2020 [cited by examiner]
US 12073638B1 · Lazar · 2024 [cited by examiner]
US 12079992B1 · Lazar · 2024 [cited by examiner]
US 12119090B1 · Kraus et al. · 2024 [cited by applicant]
US 12373950B1 · Minasian · 2025 [cited by examiner]
US 12374429B1 · Fay · 2025 [cited by examiner]
US 20100223276A1 · Al-Shameri et al. · 2010 [cited by applicant]
US 20150320319A1 · Alfano et al. · 2015 [cited by applicant]
US 20190114390A1 · Donner · 2019 [cited by examiner]
US 20200126637A1 · Xu et al. · 2020 [cited by applicant]
US 20210133976A1 · Carmi · 2021 [cited by applicant]
US 20210257044A1 · Califano · 2021 [cited by examiner]
US 20210366577A1 · Koller et al. · 2021 [cited by applicant]
US 20210374553A1 · Li et al. · 2021 [cited by applicant]
US 20220180975A1 · Regev et al. · 2022 [cited by applicant]
US 20230416307A1 · Lopez · 2023 [cited by examiner]
US 20240112447A1 · Sawada et al. · 2024 [cited by applicant]
US 20240304009A1 · Yoon et al. · 2024 [cited by applicant]
US 20260051164A1 · Nykl · 2026 [cited by examiner]
Y. Zhang, L. Yang and V. Brusic, “Automation of Gene Expression Profile Analysis in Single Cell Data,” 2020 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), Seoul, Korea (South), 2020, pp. 1329-13… [cited by examiner]
U.S. Appl. No. 18/526,729, May 27, 2025, Notice of Allowance. [cited by applicant]
Akshay Agrawal, Alnur Ali, Stephen Boyd, et al. Minimum-distortion embedding. Foundations and Trends® in Machine Learning, 14(3):211-378, 2021. [cited by applicant]
Arthur Liberzon, Chet Birger, Helga Thorvaldsdóttir, Mahmoud Ghandi, Jill P Mesirov, and Pablo Tamayo. The molecular signatures database hallmark gene set collection. Cell systems, 1(6):417-425, 2015. [cited by applicant]
Atray Dixit, Oren Parnas, Biyu Li, Jenny Chen, Charles P Fulco, Livnat Jerby-Arnon, Nemanja D Marjanovic, Danielle Dionne, Tyler Burks, Raktima Raychowdhury, et al. Perturb-seq: dissecting molecular circuits with scalab… [cited by applicant]
Aurora S Blucher, Safiye Celik, James D Jensen, James Taylor, Michael F Cuccarese, Jacob C Cooper, Jacob M Rinaldi, Carl Brooks, Michael A Statnick, Marta Fay, Nathan Lazar, Berton Earnshaw, and Imran S Haque. Poster: M… [cited by applicant]
D Michael Ando, Cory Y McLean, and Marc Berndl. Improving phenotypic measurements in high-content imaging screens. BioRxiv, p. 161422, 2017. [cited by applicant]
Gabor J Szekely. Potential and kinetic energy in statistics. Lecture Notes, Budapest Institute, 1989. [cited by applicant]
Gökcen Eraslan, Lukas M Simon, Maria Mircea, Nikola S Mueller, and Fabian J Theis. Singlecell ra-seq denoising using a deep count autoencoder. Nature communications, 10(1):1-14, 2019. [cited by applicant]
John W Tukey. Mathematics and the picturing of data. In Proceedings of the International Congress of Mathematicians, Vancouver, 1975, vol. 2, pp. 523-531, 1975. [cited by applicant]
Joseph M Replogle, Reuben A Saunders, Angela N Pogson, Jeffrey A Hussmann, Alexander Lenail, Alina Guna, Lauren Mascibroda, Eric J Wagner, Karen Adelman, Gila Lithwick-Yanai, et al. Mapping information-rich genotype-phe… [cited by applicant]
Kevin Drew, John B Wallingford, and Edward M Marcotte. hu.map 2.0: integration of over 15,000 proteomic experiments builds a global compendium of human multiprotein assemblies. Mol Syst Biol, 17(5):e10016, 2021. [cited by applicant]
Kihyuk Sohn, Honglak Lee, and Xinchen Yan. Learning structured output representation using deep conditional generative models. Advances in neural information processing systems, 28, 2015. [cited by applicant]
Krzysztof Polanski, Matthew D Young, Zhichao Miao, Kerstin B Meyer, Sarah A Teichmann, and Jong-Eun Park. Bbknn: fast batch alignment of single cell transcriptomes. Bioinformatics, 36(3):964-965, 2020. [cited by applicant]
Laleh Haghverdi, Aaron TL Lun, Michael D Morgan, and John C Marioni. Batch effects in single-cell rna-sequencing data are corrected by matching mutual nearest neighbors. Nature biotechnology, 36(5):421-427, 2018. [cited by applicant]
Leland McInnes, John Healy, and James Melville. Umap: Uniform manifold approximation and projection for dimension reduction. arXiv preprint arXiv:1802.03426, 2018. [cited by applicant]
Luana Licata, Prisca Lo Surdo, Marta Iannuccelli, Alessandro Palma, Elisa Micarelli, Livia Perfetto, Daniele Peluso, Alberto Calderone, Luisa Castagnoli, and Gianni Cesareni. Signor 2.0, the signaling network open resou… [cited by applicant]
Madalina Giurgiu, Julian Reinhard, Barbara Brauner, Irmtraud Dunger-Kaltenbach, Gisela Fobo, Goar Frishman, Corinna Montrone, and Andreas Ruepp. Corum: the comprehensive resource of mammalian protein complexes—2019. Nuc… [cited by applicant]
Marc Gillespie, Bijay Jassal, Ralf Stephan, Marija Milacic, Karen Rothfels, Andrea Senff-Ribeiro, Johannes Griss, Cristoffer Sevilla, Lisa Matthews, Chuqiao Gong, et al. The reactome pathway knowledgebase 2022. Nucleic … [cited by applicant]
Maria L Rizzo and Gábor J Székely. Energy distance. wiley interdisciplinary reviews: Computational statistics, 8(1):27-38, 2016. [cited by applicant]
Mark-Anthony Bray, Shantanu Singh, Han Han, Chadwick T Davis, Blake Borgeson, Cathy Hartland, Maria Kost-Alimova, Sigrun M Gustafsdottir, Christopher C Gibson, and Anne E Carpenter. Cell painting, a high-content image-b… [cited by applicant]
Michael F Cuccarese, Berton A Earnshaw, Katie Heiser, Ben Fogelson, Chadwick T Davis, Peter F McLean, Hannah B Gordon, Kathleen-Rose Skelly, Fiona L Weathersby, Vlad Rodic, Ian K Quigley, Elissa D Pastuzyn, Brandon M Me… [cited by applicant]
Nathan Lazar, et al. High-Resolution Genome-wide Mapping of Chromosome-arm-scale Truncations Induced by CRISPR-Cas9 Editing published in bioRxiv on Apr. 15, 2023. [cited by applicant]
Romain Lopez, Jeffrey Regier, Michael B Cole, Michael I Jordan, and Nir Yosef. Deep generative modeling for single-cell transcriptomics. Nature methods, 15(12):1053-1058, 2018. [cited by applicant]
U.S. Appl. No. 18/393,041, Feb. 23, 2024, Office Action. [cited by applicant]
Sivanandan et al. “A Pooled Cell Painting CRISPR Screening Platform Enables de novo Inference of Gene Function by Self-supervised Deep Learning”, https://www.biorxiv.org/contenU10.1101/2023.08.13.553051v3.abstract, bioR… [cited by applicant]
U.S. Appl. No. 18/392,989, Mar. 1, 2024, Office Action. [cited by applicant]
U.S. Appl. No. 18/392,989, Apr. 29, 2024, Notice of Allowance. [cited by applicant]
U.S. Appl. No. 18/526,729, Jan. 16, 2025, Office Action. [cited by applicant]
U.S. Appl. No. 18/393,041, May 3, 2024, Notice of Allowance. [cited by applicant]
U.S. Appl. No. 18/526,707, Jan. 15, 2026, Office Action. [cited by applicant]
U.S. Appl. No. 18/526,707, Apr. 14, 2026, Notice of Allowance. [cited by applicant]