This repository contains the codes for ''MIGGRI: a multi-instance graph neural network model for inferring gene regulatory network from spatial expression images''
We provides the extracted features in data/con_128.pkl.
The gene regulatory interactions is in data/regulatory_relationship.pkl
Training the model with validation by
python train.py --bidirectional
To retrain the model adding validation data into training with the best model setting on validation, execute
python retrain.py --model "src of the model"
A running example is given in main.sh.
To evaluate the model performance, execute
python test.py --model "scr of the model"