metadata

license: cc-by-nc-4.0
task_categories:
  - text-generation
language:
  - en
tags:
  - chemistry
  - biology
  - medical
pretty_name: approved_drug_target
size_categories:
  - 10K<n<100K
configs:
  - config_name: approved_drug_target
    data_files:
      - split: train
        path: approved_drug_target.json
  - config_name: uniprot_sequence
    data_files:
      - split: uniprot_seq
        path: uniprotId_sequence_2024_11_01.json

Approved Drug SMILES and Protein Sequence Dataset

This dataset provides a curated collection of approved drug Simplified Molecular Input Line Entry System (SMILES) strings and their associated protein sequences. Each small molecule has been approved by at least one regulatory body, ensuring the safety and relevance of the data for computational applications. The dataset includes 1,660 approved small molecules and their 2,093 related protein targets.

Dataset

The data comes from the following sources:

DrugBank
UniProt
ChEMBL
ZINC20

Data verification and processing

A total of 1,710 approved small molecules were retrieved from the DrugBank database, 117 of which were labeled as withdrawn. After assessment by a physician (Ali Motahharynia) and a pharmacist (Mahsa Sheikholeslami), 50 withdrawn drugs were excluded due to safety concerns, resulting in 1,660 approved small molecules. 2,116 protein targets were associated with these drugs, but 27 proteins were missing or unverified in UniProt database. These were manually replaced or verified using UniProt IDs, identical protein names, or Basic Local Alignment Search Tool (BLAST) for alignment matching, ultimately leading to 2,093 verified protein targets. Protein with UniProt ID “Q5JXX5” was deleted from the UniProt database and was therefore excluded from the dataset.

Data structure

SMILES: Contains the SMILES strings for each of the approved molecules. These SMILES were retrieved from DrugBank, ChEMBL, and ZINC20 databases.
Sequences: Contains protein sequences retrieved from UniProt database.

You can load this dataset with:

from datasets import load_dataset

dataset = load_dataset("alimotahharynia/approved_drug_target", "approved_drug_target")

You can also download the dataset directly in JSON format.

Citation

If you use this dataset in your research, please cite our paper:

@misc{sheikholeslami2024druggenadvancingdrugdiscovery,
      title={DrugGen: Advancing Drug Discovery with Large Language Models and Reinforcement Learning Feedback}, 
      author={Mahsa Sheikholeslami and Navid Mazrouei and Yousof Gheisari and Afshin Fasihi and Matin Irajpour and Ali Motahharynia},
      year={2024},
      eprint={2411.14157},
      archivePrefix={arXiv},
      primaryClass={q-bio.QM},
      url={https://arxiv.org/abs/2411.14157}, 
}