# US 2025/0095146 A1 · Utilizing machine learning and digital embedding processes to generate digital maps of biology and user interfaces for evaluating map efficacy > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2025/0095146 A1 - Document ID: US20250095146A1 - Status: Published application - Kind: A1 - Application number: 18/779,917 - Filed: 2024-07-22 - Published: 2025-03-20 - Earliest claimed: 2023-09-14 - Assignees: Recursion Pharmaceuticals, Inc. - Applicants: Recursion Pharmaceuticals, Inc. - Inventors: Nathan Henry LAZAR; Conor Austin Forsman TILLINGHAST; James Douglas JENSEN; James Benjamin TAYLOR; Berton Allen EARNSHAW; Marta Marie FAY; Renat Nailevich KHALIULLIN; Jacob Carter COOPER; Imran Saeedul HAQUE; Seyhmus GULER; Kyle Rollins HANSEN; Safiye CELIK - Web page: https://patentgrep.com/patent/US20250095146A1 ## CPC - G06V 20/698: {Matching; Classification} - G06F 3/0482: Interaction with lists of selectable items, e.g. menus - G06F 16/51: Indexing; Data structures therefor; Storage structures - G06F 16/583: using metadata automatically derived from the content - G06N 3/045: Combinations of networks - G06T 7/0012: {Biomedical image inspection} - G06T 7/35: using statistical methods - G06V 10/761: {Proximity, similarity or dissimilarity measures} - G06V 10/82: using neural networks - G06V 10/95: {structured as a network, e.g. client-server architectures} - G16B 15/00: ICT specially adapted for analysing two-dimensional [2D] or three-dimensional [3D] molecular structures, e.g. structural or functional relations or structure alignment - G16B 40/00: ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding - G16B 40/20: Supervised data analysis - G16B 50/30: Data warehousing; Computing architectures - G06T 2200/24: involving graphical user interfaces [GUIs] - G06T 2207/20081: Training; Learning - G06T 2207/20084: Artificial neural networks [ANN] - G06T 2207/30072: Microarray; Biochip, DNA array; Well plate - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL - G16B: BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY ## Abstract The present disclosure relates to systems, non-transitory computer-readable media, and methods for utilizing machine learning and digital embedding processes to generate digital maps of biology and user interfaces for evaluating map efficacy. In particular, in one or more embodiments, the disclosed systems receive perturbation data for a plurality of perturbation experiment units corresponding to a plurality of perturbation classes. Further, the systems generate, utilizing a machine learning model, a plurality of perturbation experiment unit embeddings from the perturbation data. Additionally, the systems align, utilizing an alignment model, the plurality of perturbation experiment unit embeddings to generate aligned perturbation unit embeddings. Moreover, the systems aggregate the aligned perturbation unit embeddings to generate aggregated embeddings. Furthermore, the systems generate perturbation comparisons utilizing the perturbation-level embeddings. ## Claims ### Claim 1 A computer-implemented method comprising: receiving, from a client device, a similarity request for two or more cell perturbations; generating similarity measures between the two or more cell perturbations by comparing aggregated embeddings corresponding to the two or more cell perturbations, wherein the aggregated embeddings are aggregated from perturbation experiment unit embeddings generated utilizing a machine learning model from perturbation data; and in response to the similarity request, providing, for display via the client device, a perturbation similarity map reflecting the similarity measures between the two or more cell perturbations. ### Claim 2 (depends on 1) The computer-implemented method of claim 1, wherein receiving, from the client device, the similarity request for two or more cell perturbations comprises: receiving a compound-gene similarity request between a compound perturbation and a gene perturbation; or receiving a compound-compound similarity request between a first compound perturbation and a second compound perturbation. ### Claim 3 (depends on 1) The computer-implemented method of claim 1, wherein generating the similarity measures comprises comparing a first aggregated embedding for a first perturbation and a second aggregated embedding for a second perturbation in a machine learning feature space to generate a similarity measure. ### Claim 4 (depends on 1) The computer-implemented method of claim 1, further comprising generating the perturbation experiment unit embeddings utilizing the machine learning model from the perturbation data by: capturing phenomic images of cells exposed to perturbations; and generating, utilizing the machine learning model, perturbation experiment unit embeddings from the phenomic images of the cells. ### Claim 5 (depends on 1) The computer-implemented method of claim 1, further comprising generating the perturbation experiment unit embeddings utilizing the machine learning model from the perturbation data by: generating transcriptomic profiles of cells exposed to perturbations; and generating the perturbation experiment unit embeddings from the transcriptomic profiles. ### Claim 6 (depends on 1) The computer-implemented method of claim 1, wherein providing the perturbation similarity map reflecting the similarity measures between the two or more cell perturbations for display comprises providing, for display via the client device, a heatmap comprising cells having shading that represents the similarity measures. ### Claim 7 (depends on 1) The computer-implemented method of claim 1, wherein providing the perturbation similarity map further comprises providing, for display via the client device, a confidence measure for one or more of the aggregated embeddings. ### Claim 8 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: receive, from a client device, a similarity request for two or more cell perturbations; generate similarity measures between the two or more cell perturbations by comparing aggregated embeddings corresponding to the two or more cell perturbations, wherein the aggregated embeddings are aggregated from perturbation experiment unit embeddings generated utilizing a machine learning model from perturbation data; and in response to the similarity request, provide, for display via the client device, a perturbation similarity map reflecting the similarity measures between the two or more cell perturbations. ### Claim 9 (depends on 8) The system of claim 8, further comprising instructions that, when executed by the at least one processor, cause the system to: receive a compound-gene similarity request between a compound perturbation and a gene perturbation; receive a compound-compound similarity request between a first compound perturbation and a second compound perturbation; and provide, for display via the client device, the perturbation similarity map reflecting a first similarity measure between the compound perturbation and the gene perturbation and a second similarity measure between the first compound perturbation and the second compound perturbation. ### Claim 10 (depends on 8) The system of claim 8, further comprising instructions that, when executed by the at least one processor, cause the system to generate the similarity measures by comparing a first aggregated embedding for a first perturbation and a second aggregated embedding for a second perturbation in a machine learning feature space to generate a similarity measure. ### Claim 11 (depends on 8) The system of claim 8, further comprising instructions that, when executed by the at least one processor, cause the system to generate the perturbation experiment unit embeddings utilizing the machine learning model from the perturbation data by: capturing phenomic images of cells exposed to perturbations; and generating, utilizing the machine learning model, perturbation experiment unit embeddings from the phenomic images of the cells. ### Claim 12 (depends on 8) The system of claim 8, further comprising instructions that, when executed by the at least one processor, cause the system to generate the perturbation experiment unit embeddings utilizing the machine learning model from the perturbation data by: generating transcriptomic profiles of cells exposed to perturbations; and generating the perturbation experiment unit embeddings from the transcriptomic profiles. ### Claim 13 (depends on 8) The system of claim 8, further comprising instructions that, when executed by the at least one processor, cause the system to provide the perturbation similarity map reflecting the similarity measures between the two or more cell perturbations for display by providing, for display via the client device, a heatmap comprising cells having shading that represents the similarity measures. ### Claim 14 (depends on 8) The system of claim 8, further comprising instructions that, when executed by the at least one processor, cause the system to provide the perturbation similarity map by providing, for display via the client device, a confidence measure for one or more of the aggregated embeddings. ### Claim 15 A non-transitory computer-readable medium storing instructions that, when executed by at least one processor, cause a computing device to: receive, from a client device, a similarity request for two or more cell perturbations; generate similarity measures between the two or more cell perturbations by comparing aggregated embeddings corresponding to the two or more cell perturbations, wherein the aggregated embeddings are aggregated from perturbation experiment unit embeddings generated utilizing a machine learning model from perturbation data; and in response to the similarity request, provide, for display via the client device, a perturbation similarity map reflecting the similarity measures between the two or more cell perturbations. ### Claim 16 (depends on 15) 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: receive a compound-gene similarity request between a compound perturbation and a gene perturbation; receive a compound-compound similarity request between a first compound perturbation and a second compound perturbation; and provide, for display via the client device, the perturbation similarity map reflecting a first similarity measure between the compound perturbation and the gene perturbation and a second similarity measure between the first compound perturbation and the second compound perturbation. ### Claim 17 (depends on 15) 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 generate the similarity measures by comparing a first aggregated embedding for a first perturbation and a second aggregated embedding for a second perturbation in a machine learning feature space to generate a similarity measure. ### Claim 18 (depends on 15) 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 generate the perturbation experiment unit embeddings utilizing the machine learning model from the perturbation data by: capturing phenomic images of cells exposed to perturbations; and generating, utilizing the machine learning model, perturbation experiment unit embeddings from the phenomic images of the cells. ### Claim 19 (depends on 15) 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 the perturbation similarity map reflecting the similarity measures between the two or more cell perturbations for display by providing, for display via the client device, a heatmap comprising cells having shading that represents the similarity measures. ### Claim 20 (depends on 15) 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 the perturbation similarity map by providing, for display via the client device, a confidence measure for one or more of the aggregated embeddings.