AbODE: Ab initio Antibody Design using Conjoined ODEs

The repository for AbODE: Ab initio Antibody Design using Conjoined ODEs ICML 2023 paper. For more information visit the website.

Evaluation and pre-trained models.

Prerequisites

• torchdiffeq : https://github.com/rtqichen/torchdiffeq.
• pytorch >= 1.10.0
• torch-scatter
• torch-sparse
• torch-cluster
• torch-spline-conv
• torch-geometric == 2.0.4
• astropy
• networkx
• tqdm
• Biopython

Make sure to have the correct version of each library as some errors might arise due to version-mismatch. The libraries-version of the local conda env is in env_list.txt

Datasets

Datasets are placed in the "data/" folder. Note that Structural Antibody Dataset is an evolutionary dataset in which new structures are added each week. The model has been evaluated on the same specification as other competing methods in benchmarks.

Evaluation:

Unconditional Antibody Sequence and structure generation:

python evaluation_uncond.py --cdr ${CDR-region-antibody}(1/2/3)

Conditional Antibody Sequence and structure generation:

python evaluation_cond.py --cdr ${CDR-region-antibody}(1/2/3)