README.md

---
license: cc-by-4.0
size_categories:
- 100K<n<1M
task_categories:
- image-classification
tags:
- biology
- drug
- cells
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
dataset_info:
  features:
  - name: image
    dtype:
      array3_d:
        dtype: uint8
        shape:
        - 512
        - 512
        - 6
  - name: site_id
    dtype: string
  - name: well_id
    dtype: string
  - name: cell_type
    dtype: string
  - name: experiment
    dtype: string
  - name: plate
    dtype: int32
  - name: well
    dtype: string
  - name: site
    dtype: int32
  - name: well_type
    dtype:
      class_label:
        names:
          '0': treatment
          '1': positive_control
          '2': negative_control
  - name: sirna
    dtype: string
  - name: sirna_id
    dtype: int32
  - name: embeddings
    sequence: float32
    length: 128
  splits:
  - name: train
    num_bytes: 213139738276
    num_examples: 81224
  - name: test
    num_bytes: 116210798412
    num_examples: 44286
  download_size: 0
  dataset_size: 329350536688
---

[![DOI](https://zenodo.org/badge/DOI/10.48550/arXiv.2301.05768.svg)](https://doi.org/10.48550/arXiv.2301.05768)

# RxRx1: A Dataset for Evaluating Experimental Batch Correction Methods

**Homepage**: https://www.rxrx.ai/rxrx1 \
**Publication Date**: 2019-06 \
**License**: [Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)](https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode) \
**Citation**:
```bibtex
@misc{sypetkowski2023rxrx1,
  title         = {RxRx1: A Dataset for Evaluating Experimental Batch Correction Methods},
  author        = {Maciej Sypetkowski and Morteza Rezanejad and Saber Saberian and Oren Kraus and John Urbanik and James Taylor and Ben Mabey and Mason Victors and Jason Yosinski and Alborz Rezazadeh Sereshkeh and Imran Haque and Berton Earnshaw},
  year          = {2023},
  eprint        = {2301.05768},
  archiveprefix = {arXiv},
  primaryclass  = {cs.CV}
}
```

## Description

High-throughput screening techniques are commonly used to obtain large quantities of data in many fields of biology. It is well known that artifacts arising from variability in the technical execution of different experimental batches within such screens confound these observations and can lead to invalid biological conclusions. It is therefore necessary to account for these batch effects when analyzing outcomes. In this paper we describe RxRx1, a biological dataset designed specifically for the systematic study of batch effect correction methods. The dataset consists of 125,510 high-resolution fluorescence microscopy images of human cells under 1,138 genetic perturbations in 51 experimental batches across 4 cell types. Visual inspection of the images alone clearly demonstrates significant batch effects. We propose a classification task designed to evaluate the effectiveness of experimental batch correction methods on these images and examine the performance of a number of correction methods on this task. Our goal in releasing RxRx1 is to encourage the development of effective experimental batch correction methods that generalize well to unseen experimental batches.