eachanjohnson's picture
Initial upload (#1)
2b62bea verified
metadata
license: mit
task_categories:
  - text-classification
  - text2text-generation
  - translation
  - zero-shot-classification
tags:
  - chemistry
  - biology
  - SMILES
  - benchmark
size_categories:
  - 1k<n<10k
pretty_name: Biogen ADME (public data)
configs:
  - config_name: full
    data_files: biogen-adme.csv.gz
  - config_name: scaffold-split
    data_files:
      - split: train
        path: biogen-adme_train.csv.gz
      - split: test
        path: biogen-adme_test.csv.gz

Biogen ADME dataset (public data)

Data from Fang et al., Prospective Validation of Machine Learning Algorithms for Absorption, Distribution, Metabolism, and Excretion Prediction: An Industrial Perspective, available from the GitHub repositiory. We used schemist (which in turn uses RDKit) to add molecuar weight, Murcko scaffold, Crippen cLogP, and topological surface area, and to generate scaffold splits.

Dataset Details

From the original README:

To benefit the broader computational chemistry community and improve the > quality and diversity of public-domain ADME data sets we have disclosed a > collection of 3521 diverse compounds selected from commercially available > compound libraries (i.e. Enamine, eMolecules, WuXi LabNetwork, Mcule) and > tested them against our internal six ADME in vitro assays described in this > study using the same experimental conditions as of our in-house datasets.

Dataset Description

  • Curated by: Biogen
  • License: MIT

Dataset Sources

Uses

Benchmarking chemical property prediction models.

Dataset Structure

The train-test splits are generated by scaffold.

The column headings of the data are:

  • SMILES: Original SMILES string, as in the original data release in the GitHub repositiory
  • smiles: Canonicalized SMILES string
  • id: Numeric structure identifier
  • inchikey: Unique structure identifier
  • scaffold: Murcko scaffold
  • mwt: Molecular weight
  • clogp: Crippen LogP
  • tpsa: Calculated topological polar surface area.

The following columns are ADME properties:

  • log_hlm: human liver microsomal (HLM) stability (Clint, mL/min/kg)
  • log_mdr1_mdck_er: MDR1-MDCK efflux ratio
  • log_solubility: solubility at pH 6.8 (ug/mL)
  • log_plasma_protein_binding_human: human plasma protein binding (hPPB) percent unbound
  • log_plasma_protein_binding_rat: rat plasma protein binding (rPPB) percent unbound
  • log_rlm: rat liver microsomal (RLM) stability (Clint, mL/min/kg)

Dataset Creation

Curation Rationale

To make the Biogen ADME dataset readily available with light preprocessing.

Data Collection and Processing

Additional properties and scaffold splits were calculated using schemist, a tool for processing chemical datasets.

Who are the source data producers?

Biogen

Personal and Sensitive Information

None

Citation

BibTeX:

@article{doi:10.1021/acs.jcim.3c00160,
    author = {Fang, Cheng and Wang, Ye and Grater, Richard and Kapadnis, Sudarshan and Black, Cheryl and Trapa, Patrick and Sciabola, Simone},
    title = {Prospective Validation of Machine Learning Algorithms for Absorption, Distribution, Metabolism, and Excretion  Prediction: An Industrial Perspective},
    journal = {Journal of Chemical Information and Modeling},
    volume = {63},
    number = {11},
    pages = {3263-3274},
    year = {2023},
    doi = {10.1021/acs.jcim.3c00160},
    note = {PMID: 37216672},
    URL = {https://doi.org/10.1021/acs.jcim.3c00160},
    eprint = {https://doi.org/10.1021/acs.jcim.3c00160}
}

Dataset Card Contact

@eachanjohnson