Info This dataset contains the supplementary materials for "Improving diffusion-based protein backbone generation with global-geometry-aware latent encoding". For source code and detailed instructions on usage, please refer to our github . Supplementary data weights.tar.gz The trained model weights used in the paper. dataset.zip CATH-60 Dataset used in the paper. In the notebook directory of our github , we provide an example on encoding and visualize it with our trained encoder. design.zip The 21 novel mainly-beta designs selected for experiment validation. Along with the generated backbone, we also provide the prediction results from AlphaFold and ESMFold. benchmark_sample.zip Sampled backbones used for all benchmark experiment (All methods and variants included). evaluation.tar.gz Precomputed CATH reference data for coverage metric computation. Need to be downloaded for using evaluation scripts.