| ## CellCLIP - Learning Perturbation Effects in Cell Painting via Text-Guided Contrastive Learning | |
| CellCLIP, a cross-modal contrastive learning framework for HCS data. CellCLIP leverages pre-trained image encoders coupled with a novel channel encoding scheme to better capture relationships between different microscopy channels in image embeddings, along with natural language encoders for repre senting perturbations. Our framework outperforms current open-source models, demonstrating the best performance in both cross-modal retrieval and biologically meaningful downstream tasks while also achieving significant reductions in computation time. | |
| * [Paper](https://arxiv.org/pdf/2506.06290) | |
| * [Github](https://github.com/suinleelab/CellCLIP/tree/main) | |
| This repository contains model checkpoints for CellCLIP trained with | |
| * Cell painting encodings: Image embeddings extracted using DINOv2-Giant and projected to a feature dimension of 1536. | |
| * Perturbation encodings: Text embeddings generated using BERT as the text encoder. | |
| ## Citation | |
| ``` | |
| @article{lu2025cellclip, | |
| title={CellCLIP--Learning Perturbation Effects in Cell Painting via Text-Guided Contrastive Learning}, | |
| author={Lu, Mingyu and Weinberger, Ethan and Kim, Chanwoo and Lee, Su-In}, | |
| journal={arXiv preprint arXiv:2506.06290}, | |
| year={2025} | |
| } | |
| ``` | |