The codes for the paper: "BiU-Net: A Biologically Informed U-Net for Genotype Imputation"

Before training/testing/benchmarking, please make sure the dataset has been prepared:

1. Genotype data of specific population have undergone QC, and splitted into train/val/test by given proportion.
2. Test sets have been masked with given missing ratio and random states
3. Segmentation has been performed.

All datasets except the in-house LOS dataset, are located at: https://github.com/learnslowly/data