---
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]](#1), part of MoleculeNet [[2]](#2) benchmark. It is intended to be used through 
[scikit-fingerprints](https://github.com/scikit-fingerprints/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
<a id="1">[1]</a> 
AIDS Antiviral Screen Data
https://wiki.nci.nih.gov/display/NCIDTPdata/AIDS+Antiviral+Screen+Data

<a id="2">[2]</a> 
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