MFFHA-DTI Dataset. Zhao et al.

The MFFHA-DTI dataset contains three datasets, Human, C.elegans and interaction type dataset. The public datasets Human and C.elegans are used to predict the probability of drug target interaction. They are universal and effective in drug research and development. These two datasets are composed of smiles structural formula, protein sequence, and 1 and 0 Tags of whether they interact. They contain highly reliable negative DTI. In short, it is a balanced DTI dataset, in which the ratio of positive samples to negative samples (positive: negative) is about 1:1. The newly constructed dataset is used to predict the specific type of drug target interaction. The data is filtered from the 2024-04 version of all drugs dataset in the drugbank database. The smiles structural formula of the drug, the corresponding protein sequence of a single protein that reacts, and the type of interaction between them are selected. Then the pdb file of the corresponding protein is obtained from alphafolddb website according to the UniProt ID in the all drugs dataset, and then the data obtained from the drugbank database is mapped according to the ID.