Taba is a computational tool for the development of machine-learning models to predict the affinity between ligands and proteins. Taba uses information extracted from the three-dimensional structures of protein-ligand complexes.
Computational methods to evaluate protein-ligand interactions exert great beneficial impact on the early stages of drug-design and development. Although much development in this field has been achieved, there is room for further progress in the creation of protein-targeted scoring functions for calculation of ligand binding affinity. It was with this in mind that we propose here a new computational tool to create machine- learning models to calculate ligand-binding affinity. The computational tool is called Taba, an acronym for tool to analyze the binding affinity.
Taba is an open source software and makes use of algorithms of supervised machine learning such as least absolute shrinkage and selection operator (Lasso) and elastic net to create a scoring function aimed to be used for a specific protein family. Taba was developed using Python programming language and makes use of scientific computing libraries such as NumPy, SciPy, Matplotlib, and Scikit-learn. Taba calculates the average interatomic distances between pairs of atoms involving protein and ligand using atomic coordinates stored in protein data bank (PDB) files.
Protein-ligand as a mass-spring system. We used the atomic coordinates for the complex CDK2-roscovitine (PDB: 2A4L) (De Azevedo et al., 1997).
Attention: If you already have a version of Taba installed, before making an update it is important to save the current experiment in Manage Experiments.
Main window of Taba
To test Taba you can use the codes of this structure:
1AZ1, 1MAR, 2DUX, 2DUZ, 2DV0, 2FZ8, 2FZ9, 2HV5, 2HVN, 2HVO, 2INE, 2INZ, 2IPW, 2IQ0, 2IQD, 2IS7, 2ISF, 2PD5, 2PDB, 2PDF, 2PDI, 2PDM, 2PDX, 3LEN, 3M0I, 3MB9, 3T42, 3V35
Website figure: http://www.dicasdamel.com.br