---
license: mit
tags:
  - spatial-transcriptomics
  - graph-transformer
  - gene-expression
  - finetuned
  - mouse-stroke
  - pytorch
language:
  - en
library_name: transformers
pipeline_tag: feature-extraction
---

# SpatialGT Finetuned Model - Mouse Stroke (PT)

## Model Description

This is the **finetuned checkpoint** of SpatialGT on mouse stroke PT (photothrombotic stroke) spatial transcriptomics data.

This model is specifically finetuned for the mouse stroke perturbation simulation case study, trained on the PT1-1 slice.

## Model Details

- **Base Model**: SpatialGT Pretrained
- **Finetuning Data**: Mouse stroke PT 4 slices (Visium)
- **Finetuning Strategy**: Full finetuning (8 transformer layers unfrozen)
- **Epochs**: 100
- **Learning Rate**: 1e-4

## Usage

```python
import torch
from pretrain.model_spatialpt import SpatialNeighborTransformer
from pretrain.Config import Config

# Load configuration
config = Config()

# Initialize model
model = SpatialNeighborTransformer(config)

# Load finetuned weights
from safetensors.torch import load_file
state_dict = load_file("model.safetensors")
model.load_state_dict(state_dict)

model.eval()
```

## Intended Use

This model is intended for:
- Reconstructing gene expression in stroke-affected mouse brain tissue
- Simulating perturbation effects in ischemic regions (ICA, PIA)
- Comparative analysis with Sham (control) model

## Files

- `model.safetensors`: Model weights in safetensors format
- `training_args.bin`: Training arguments

## Related Models

- [SpatialGT-Pretrained](https://huggingface.co/Bgoood/SpatialGT-Pretrained): Base pretrained model
- [SpatialGT-MouseStroke-Sham](https://huggingface.co/Bgoood/SpatialGT-MouseStroke-Sham): Finetuned on Sham (control) slice

## Citation

If you use this model, please cite our paper (details to be added upon publication).

## License

MIT License