File size: 1,554 Bytes
656047d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 |
---
library_name: perturblab
tags:
- biology
- genomics
- scfoundation
- foundation-model
license: apache-2.0
base_model: biomap-research/scFoundation
---
# scfoundation-rde
## Model Description
This is the **RDE (Reaction-Diffusion Embedding)** model from scFoundation. It provides specialized embeddings optimized for specific downstream tasks.
Model weights were originally from the [biomap-research/scFoundation](https://github.com/biomap-research/scFoundation) repository and have been re-uploaded here for ease of use with the `perturblab` library.
## Model Details
- **Model Type**: RDE embedding model
- **Architecture**: xTrimoGene with MAE (Masked Autoencoder), Performer/Transformer modules
- **Parameters**: 100M parameters
- **Training Data**: 50M+ human single-cell transcriptomics data
- **Input**: Single-cell or bulk RNA-seq expression data (19,264 fixed genes)
- **Output**: RDE-optimized embeddings
## Source
- **Original Repository**: [biomap-research/scFoundation](https://github.com/biomap-research/scFoundation)
- **Paper**: [Large Scale Foundation Model on Single-cell Transcriptomics](https://www.nature.com/articles/s41592-024-02305-7) (_Nature Methods_, 2024)
## Usage
```python
from perturblab.model.scfoundation import scFoundationModel
# Load model
model = scFoundationModel.from_pretrained('scfoundation-rde', device='cuda')
# Generate embeddings
embeddings = model.predict_embedding(
adata,
output_type='cell',
pool_type='max'
)
```
## Note
Intended for internal use with the PerturbLab framework.
|