Pre-trained deep models for protein contact/distance/orientation/SS/ACC and backbone torsion angle prediction

This dataset contains two sets of deep ResNet models for RaptorX-3DModeling. One set of models is used for the prediction of backbone phi/psi angles, secondary structure and solvent accessibility and the other set of models is used for the prediction of inter-atom contact, distance and orientation probability distribution. The RaptorX-3DModeling package itself is available at https://github.com/j3xugit/RaptorX-3DModeling/ . Compared to the deep model set at https://data.mendeley.com/datasets/5dj63g5xdv/1, this dataset has 4 more deep models for secondary structure and solvent accessibility prediction.

Meanwhile. RXDeepModels4DistOri-FM.tar.gz is used for inter-residue (atom) contact/distance/orientation prediction and RXDeepModels4Property.tar.gz is used for phi/psi angle, secondary structure (3-class and 8-class) and solvent accessibility prediction.