This supplementary material provides detailed information about the deep learning model used in the RoseTTAFold protein structure prediction method. The model is designed to predict protein structures and interactions using a 3-track network, which includes multiple sequence alignments (MSAs), pair features, and 3D coordinates. The materials and methods section covers the initial embedding of MSAs, processing MSA features via self-attention, updating pair features with coevolution signals, refining pair features via row and column-wise self-attention, and updating MSA features based on structure information encoded in pair features. The model also includes an initial 3D structure prediction using a Graph Transformer architecture and refinement using an SE(3)-Transformer. The training details, including the extended trRosetta training set and hyperparameters, are provided. Additionally, the material includes a tutorial on how to install and use the RoseTTAFold method, as well as examples of successful applications in molecular replacement calculations, GPCR modeling, and hetero-complex structure prediction.This supplementary material provides detailed information about the deep learning model used in the RoseTTAFold protein structure prediction method. The model is designed to predict protein structures and interactions using a 3-track network, which includes multiple sequence alignments (MSAs), pair features, and 3D coordinates. The materials and methods section covers the initial embedding of MSAs, processing MSA features via self-attention, updating pair features with coevolution signals, refining pair features via row and column-wise self-attention, and updating MSA features based on structure information encoded in pair features. The model also includes an initial 3D structure prediction using a Graph Transformer architecture and refinement using an SE(3)-Transformer. The training details, including the extended trRosetta training set and hyperparameters, are provided. Additionally, the material includes a tutorial on how to install and use the RoseTTAFold method, as well as examples of successful applications in molecular replacement calculations, GPCR modeling, and hetero-complex structure prediction.