Ames Mutagenicity Predictions - Random Forest
Service is currently down to license upgrades
This page uses random forest model
to predict
a structure (specified in SMILES format) is mutagenic or not (according to the
Ames test), based on
166 bit MACCS fingerprints. The model is based on a 4337 compound dataset studied by Kazius et al.
(J. Med. Chem, 2005, 48(1), 312-320), which was
divided approximately equally (1.2:1) between mutagens and nonmutagens.
OOB error was 15.39% and test set error was 15.32%
Note, this page is a proof-of-concept of model deployment using
R web services coupled to a remote R engine. The model development did not investigate other features
and only really tuned the number of descriptors searched at each split.
Paste SMILES, one to a line