
by Baker Lab
ProteinMPNN is a deep learning code base for designing protein sequences that fit a given backbone structure. You run it from the command line on PDB files.
Web appFree
Step 1: Clone the repository and install Python, PyTorch and NumPy, for example in a conda environment.
Step 2: Provide a PDB file or parsed PDB data as input.
Step 3: Run protein_mpnn_run.py with flags for chains, fixed positions, sampling temperature and sequences per target.
Step 4: Read the generated sequences and scores from the output folder.
Step 5: Look at the example scripts for fixed positions, symmetry, bias and score-only runs.
Clone the GitHub repository, install Python, PyTorch and NumPy, and run protein_mpnn_run.py with your input PDB. Colab notebooks are also included.
Yes. You can specify which chains to design and provide fixed positions, and the examples show both.
Yes. A score-only mode scores input backbone and sequence pairs, including sequences loaded from a FASTA file.
DeepCellDeep-learning platform and models for cell segmentation, tracking, and quantitative analysis of microscopy images