Update README.md

README.md

@@ -11,4 +11,32 @@ tags:
 pretty_name: MoleculeNet BBBP
 size_categories:
 - 1K<n<10K
----
+---
+
+# MoleculeNet BBBP
+
+BBBP (Blood-Brain Barrier Penetration) 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 blood-brain barrier penetration (barrier permeability) of small drug-like molecules.
+
+| **Characteristic** | **Description** |
+|:------------------:|:---------------:|
+|        Tasks       |        1        |
+|      Task type     |  classification |
+|    Total samples   |       2039      |
+|  Recommended split |     scaffold    |
+| Recommended metric |      AUROC      |
+
+## References
+<a id="1">[1]</a> 
+Ines Filipa Martins et al.
+"A Bayesian Approach to in Silico Blood-Brain Barrier Penetration Modeling"
+J. Chem. Inf. Model. 2012, 52, 6, 1686–1697
+https://pubs.acs.org/doi/10.1021/ci300124c
+
+<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