Downloads
Shared for science
This is a repository of downloadable datasets, GitHub links to code repositories, executable Colab Notebooks, and other research write-ups from the Chowdhury Lab. All software is licensed under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
RealKcat - Catalytically Aware Enzyme Kcat predictor, trained on KinHub-27k (Manually-curated Enzyme Parameter Database)
The enzyme kinetic parameters kcat and Km were systematically curated from BRENDA and SABIO-RK (as of May 2024), resulting in a initial dataset of 30,442 kcat and 44,615 Km entries. It is annotated with substrate SMILES from PubChem.
KinHub-27k has:
- 26,244 entries with both kcat and Km values
- 16,000 wild-type, 11,178 mutant entries
- 1:1 negative to positive data
- Manually curated from 2,158 scientific PMIDs
- A thorough and manual curation resolved >10k inconsistencies in BRENDA and SABIO-RK
RealKcat (trained on KinHub-27k) is the first-of-its kind catalytically aware kinetic predictor of enzymatic reactions, which is sensitive to mutations at the active center - and the first model to be able to predict zero kcat value if the catalytic apparatus of an enzyme is mutated.
RC-Hydrolase
RC-Hydrolase is the first reactive center database for three-dimensional structural data of catalytic/active sites and neighboring allosterically regulated regions of single and multi-chain enzyme molecules listed in the Protein Data Bank (PDB). This database enables comparison and visualizing structurally similar, yet functionally different active pockets, from desired host organisms thereby offering versatile starting scaffolds to design multi-functional enzymes for industrial biotechnology.
RC-Hydrolase contains:
- 11859 Reactive Centers
- 10 EC Classes
- 643 Organisms
- 2841 Ligands
HuMEnz
Human disease-linked metalloenzyme/ metalloprotein database (HumEnz-1.0) contains sequence, structure, and function information.
