iGWAS is a framework for extracting phenotypes from images and doing GWAS.
The code for training segmentation network is at segmentation.
The code for training quality assessment network is at quality_assessment.
The code for training embedding network is at embedding.
The name of the dataset used for training segmentation network can be found in segmentation/prepare_datasets.py, the dataset used for training embedding network can be downloaded at https://www.kaggle.com/c/diabetic-retinopathy-detection/data. The dataset used for training quality assessment network was manually created using the TKinter program quality_assessment/binary_classification_tool.py.
The GWAS folder contains wrapper functions for BOLT-LMM and Plink, and some helper functions to do additional analysis.
The locuszoom_plots folder contains all the locuszoom plots and the script to download figures from the locuszoom website.
Ziqian Xie, Tao Zhang, Sangbae Kim, Jiaxiong Lu, Wanheng Zhang, Cheng-Hui Lin, Man-Ru Wu, Alexander Davis, Roomasa Channa, Luca Giancardo, Han Chen, Sui Wang, Rui Chen, Degui Zhi. iGWAS: image-based genome-wide association of self-supervised deep phenotyping of human medical images. Submitted 2021