molclaw-pack-sidechains

You can predict full-atom sidechains from a backbone PDB by providing the backbone file as input; the tool applies AttnPacker to generate a complete output PDB with accurately placed sidechain conformations.

Full-atom sidechain packing is needed when preparing protein structures for downstream modeling workflows like docking setup, molecular dynamics initialization, or virtual screening preprocessing where accurate sidechain placement is required.

Yes, the sidechain prediction process supports chunked inference specifically for long proteins, along with device selection such as choosing cuda:0, to manage computational resources during structure preparation.

Yes, the sidechain packing includes optional rotamer post-processing with configurable optimization iterations and steric weights to reduce manual rotamer adjustment and minimize steric clashes in the generated structures.

The best way to save generated structures is by using the mandatory server_file_to_base64 download step, which ensures the output PDB and run directory files are saved locally for verification and reproducibility.