Phenomics Imaging · recursionpharma
rxrx3
High-content microscopy dataset covering cellular perturbations in human cells for drug discovery researchers.
What rxrx3 is
RxRx3 is a phenomics imaging dataset developed by Recursion Pharmaceuticals (recursionpharma) to support image-based drug discovery. It provides high-content microscopy imagery documenting cellular responses across human cell lines subject to both genetic and chemical perturbations.
The dataset integrates cellular images with corresponding text annotations across biology, medicine, and chemistry. It offers a large-scale collection of phenotypic observations to help researchers analyse and model cellular states under systematic experimental interventions.
How it was made
The dataset captures cellular phenotypes across human cell lines exposed to genetic modifications and chemical treatments using high-content microscopy. These microscopic images are paired with descriptive text annotations and formatted into Parquet and optimized-Parquet files designed for interoperability with data processing libraries such as Polars, Dask, and Datasets.
Key facts
- Category
- Phenomics Imaging
- Fields
- Biology, medicine, chemistry
- Modalities
- Image, text
- Dataset size
- 10 million to 100 million entries
- Data formats
- Parquet, optimized-parquet
What you can use it for
- Image-based drug discovery
- Screening morphological responses of human cell lines to chemical perturbations to support compound discovery and characterisation.
- Genetic perturbation analysis
- Evaluating phenotypic cellular changes caused by targeted genetic interventions.
- Multi-modal phenomics modelling
- Training computational models using paired high-content microscopy images and text annotations across biological and chemical contexts.
Explore the data
This dataset is gated
Its authors require their terms to be accepted before the data can be read. Accept them on the original dataset page. The platform does not accept terms on your behalf, so the rows are not shown here.Strengths and limits
Good at
- Combines both genetic and chemical perturbations across human cell lines in one microscopy resource.
- Contains tens of millions of records pairing multi-channel imagery with text annotations.
- Stored in Parquet and optimized-Parquet formats compatible with Polars, Dask, and Hugging Face Datasets.
Questions
Can I look inside rxrx3 without downloading it?
Not here: its authors gate it behind terms each user must accept on the original page. Its description and statistics are still shown.
What licence is rxrx3 under?
Its authors declare other. Read the terms on the original page before reusing the data.
What data modalities are included in RxRx3?
RxRx3 contains high-content microscopy images paired with text annotations.
What types of perturbations are covered in the dataset?
The dataset spans both chemical and genetic perturbations tested in human cell lines.
Which software libraries support the RxRx3 data files?
The dataset is provided in Parquet and optimized-Parquet formats supported by Datasets, Dask, Polars, and MLCroissant.
How to cite it
arXiv:2301.05768
Written from rxrx3's own dataset card. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.