metadata

license: unknown
task_categories:
  - tabular-classification
  - graph-ml
  - text-classification
tags:
  - chemistry
  - biology
  - medical
pretty_name: MoleculeNet HIV
size_categories:
  - 10K<n<100K
configs:
  - config_name: default
    data_files:
      - split: train
        path: hiv.csv

MoleculeNet HIV

HIV dataset [1], part of MoleculeNet [2] benchmark. It is intended to be used through scikit-fingerprints library.

The task is to predict ability of molecules to inhibit HIV replication.

Characteristic	Description
Tasks	1
Task type	classification
Total samples	41127
Recommended split	scaffold
Recommended metric	AUROC

References

[1] AIDS Antiviral Screen Data https://wiki.nci.nih.gov/display/NCIDTPdata/AIDS+Antiviral+Screen+Data

[2] Wu, Zhenqin, et al. "MoleculeNet: a benchmark for molecular machine learning." Chemical Science 9.2 (2018): 513-530 https://pubs.rsc.org/en/content/articlelanding/2018/sc/c7sc02664a