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.

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.

Abstract: The present disclosure relates to systems, non-transitory computer-readable media, and methods for deducing information for mechanism of actions (MOAs) utilizing digital signals from cell representations within a shared feature space. In particular, the disclosed systems can deduce (or predict) MOAs by generating MOA representations with corresponding detection confidence scores that indicate whether cell representations in a MOA representation provide a meaningful signal to predict the MOA. Indeed, the disclosed systems can determine a cluster of cell representation embeddings (in the shared feature space) based on annotated cell representation embeddings corresponding to a known MOA to generate an MOA representation. Furthermore, the disclosed systems can utilize MOA representations, within the shared feature space, to predict MOAs for a query cell representation (of a perturbation).

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.

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.

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. For example, the disclosed systems can generate a combined phenomic-transcriptomic map from embedding perturbation data via a machine learning model and filtering, aligning, aggregating, and relating the embeddings to generate transcriptomic comparisons. Additionally, the disclosed systems can embed phenomic perturbation data via a machine learning model and filtering, aligning, aggregating, and relating the phenomic perturbation embeddings to generate phenomic perturbation comparisons. Furthermore, the disclosed systems can utilize transcriptomic comparisons determined from aggregated transcriptomic embeddings and phenomic embedding comparisons determined from aggregated phenomic perturbation embeddings to generate combined phenomic-transcriptomic maps of biology.

Abstract: The present disclosure relates to systems, non-transitory computer-readable media, and methods for training and utilizing generative machine learning models to generate embeddings from phenomic images (or other microscopy representations). For example, the disclosed systems can train a generative machine learning model (e.g., a masked autoencoder generative model) to generate predicted (or reconstructed) phenomic images from masked version of ground truth training phenomic images. In some cases, the disclosed systems utilize a momentum-tracking optimizer while reducing a loss of the generative machine learning model to enable efficient training on large scale training image batches. Furthermore, the disclosed systems can utilize Fourier transformation losses with multi-stage weighting to improve the accuracy of the generative machine learning model on the phenomic images during training.

Abstract: The present disclosure relates to systems, non-transitory computer-readable media, and methods that implement a compound exploration initiation system. Indeed, in one or more implementations, the disclosed systems identify, from processed biological representations, a predicted biological relationship for an anchor compound or an anchor gene. Further, in one or more implementations, the disclosed systems generate digital text prompts that include text rating instructions for a language machine learning model from the predicted biological relationship. To illustrate, the disclosed systems generate rating metrics according to the text rating instructions utilizing the language machine learning model and combines the rating metrics to generate a program rating for the anchor gene or the anchor compound for initiating one or more compound exploration programs.

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.

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.

Abstract: The present disclosure relates to systems, non-transitory computer-readable media, and methods for training and utilizing generative machine learning models to generate embeddings from phenomic images (or other microscopy representations). For example, the disclosed systems can train a generative machine learning model (e.g., a masked autoencoder generative model) to generate predicted (or reconstructed) phenomic images from masked version of ground truth training phenomic images. In some cases, the disclosed systems utilize a momentum-tracking optimizer while reducing a loss of the generative machine learning model to enable efficient training on large scale training image batches. Furthermore, the disclosed systems can utilize Fourier transformation losses with multi-stage weighting to improve the accuracy of the generative machine learning model on the phenomic images during training.

Abstract: The present disclosure relates to systems, non-transitory computer-readable media, and methods for training and utilizing generative machine learning models to generate embeddings from phenomic images (or other microscopy representations). For example, the disclosed systems can train a generative machine learning model (e.g., a masked autoencoder generative model) to generate predicted (or reconstructed) phenomic images from masked version of ground truth training phenomic images. In some cases, the disclosed systems utilize a momentum-tracking optimizer while reducing a loss of the generative machine learning model to enable efficient training on large scale training image batches. Furthermore, the disclosed systems can utilize Fourier transformation losses with multi-stage weighting to improve the accuracy of the generative machine learning model on the phenomic images during training.

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.

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.