Deepinteract

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

1. [Ensure PDB codes match for input pairs](https://github.com/BioinfoMachineLearning/DeepInteract/commit/f301dcf076bded481a43cf517e945a2e73b19887)
2. [Restore original chain IDs to EVCoupling input chain DataFrames](https://github.com/BioinfoMachineLearning/DeepInteract/commit/6d885b4d419c528e8f3eae0c5eefb34f43223c1e)
3. [Install missing chain IDs into input PDB files](https://github.com/BioinfoMachineLearning/DeepInteract/commit/4824011ae358b47f9ea75d18278a764becfa1941)
4. [Halt execution of the inference pipeline if either input PDB is not found](https://github.com/BioinfoMachineLearning/DeepInteract/commit/d246f867190d911cff3804747b4120e589ce876b)

This release includes a small bug fix in graph construction.

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

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)](https://github.com/BioinfoMachineLearning/DeepInteract/commit/e89d9467ba4899d43efe3d01192bac7f6e93b174)

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

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)