1.1.0

• This release provides full support for the Docking Benchmark 5 (DB5) dataset within the DeepInteract training, fine-tuning, and testing pipeline.
• This release also adds DB5 download links to reference the preprocessed version of the DB5 dataset available on Zenodo.

1.0.9

This release includes a small bug fix in graph construction.

1.0.8

This release includes a few additions and updates:

1. Inclusion of top-k recall metric in our training pipeline
2. Addition of DB5 fine-tuned PyTorch LighntingModule checkpoint
3. Updates to Zenodo links to accommodate new Zenodo dataset version

1.0.7

This release includes a small feature enhancement:

1. Allow features for multiple chains in a single ligand/receptor to be postprocessed (i.e., add multimer feature postprocessing support)

1.0.5-1.0.6

This release of DeepInteract includes 4 enhancements/bug fixes inside the Docker inference pipeline:

1. Ensure PDB codes match for input pairs
2. Restore original chain IDs to EVCoupling input chain DataFrames
3. Install missing chain IDs into input PDB files
4. Halt execution of the inference pipeline if either input PDB is not found

1.0.4

This release of DeepInteract includes 1 enhancement/bug fix inside the Docker inference pipeline:

1. Install sorting and hard-indexing logic to ensure PSAIA feature DataFrames are collected and used in the correct order during inference

1.0.3

This release of DeepInteract includes 1 enhancement inside the Docker inference pipeline:

1. Recover missing PDB chain IDs in the data preprocessing pipeline (prior to inference)

1.0.2

This release of DeepInteract addresses 3 bugs and removal of deprecated logic inside the Docker inference pipeline:

1. Allow input PDB filenames to use any naming convention
2. Ensure that feature imputation occurs prior to each prediction by making the feature imputation function name unique
3. Fix file not found error for complex.dill
4. Remove deprecated use_dgl argument and its associated logic

1.0.1

This release marks the public unveiling of DeepInteract.