Tree-Based Interpretable Machine Learning Models for GxE Prediction

This is a Python implementation of our paper:

Interpretable machine learning uncovers complex, interacting traits associated with maize yield across diverse environments

Getting started

Requirements

• Python 3.8.4
• scikit-learn 1.3.0
• conda

Installation

Clone repository:

git clone https://github.com/AIBreeding/XAI.git

Create environment:

conda create -n GxE python=3.8.4
conda activate GxE

Install packages:

cd GxE
conda install --yes --file requirements.txt

Follow the instructions in data directory to get dataset files.

Follow the instructions in model/G2Pmodel to get pre-trained models.

Usage

Hyperparameter optimization

• python Start-AutoHPO.py

10-fold cross-validation data partitioning

• python kfolds.py

Model training and independent prediction

• python Start-Basic_model.py

Basic model stacking operation

• python stacking.py

Model Interpretation and Visualization

• Please execute the XAI.ipynb script