A number of services to access OCHEM using Swagger UI were developed as part of the Implementation Challenge of the https://openrisknet.org/ project.
They API is accessible at http://rest.ochem.eu
The tools provide an access to several models, namely physico-chemical properties (logP, melting point, water solubility), biological properties (BCF - bio-concentration factor, the Human Ether-a-go-go-related Gene (hERG) channel blockage, CYP1A2 inhibition) as well as toxicity prediction (multi-learning model), AMES and Lowest Effect Level (LOEL) as well as environmental toxicity (T. pyriformis). All models are part of the OCHEM platform and for each model there is a link to its reference and description in the respective publication at the OCHEM platform. The provided model interface allows an interested user to submit a SMILES of the analyzed compounds and retrieve prediction results. The predictions are performed on the OCHEM web site.
When predicting a model, the user has to provide its ID as well as valid SMILES.
Model ID is an integer value and list of models is accessible as, e.g.
curl -X GET "http://rest.ochem.eu/endpoint"
{
"AMES toxicity" : 1,
"CYP450 inhibition" : 2,
"Toxicity against T. Pyriformis" : 3,
"CYP1A2 inhibition" : 159,
"LEL - Lowest effect dose" : 174,
"BCF - Bioconcentration factor" : 8,
"Melting Point" : 501,
"logP - ALOGPS2.1" : 535,
"logS - ALOGPS2.1 solubility in water" : 536,
"hERG K+ Channel Blocking" : 628,
"In vivo toxicities" : 635
}
The respective IDS of models are provided after each model. To make predictions, you can use, e.g.
curl -X GET "http://rest.ochem.eu/predict?MODELID=1&SMILES=Cc1ccccc1"
where 1 is the model id for AMES model. The AMES model is also accessible as http://ochem.eu/model/1
The prediction is:
curl -X GET "http://rest.ochem.eu/predict?MODELID=1&SMILES=Cc1ccccc1"
{
"MODEL:" : "http://ochem.eu/model/1",
"SMILES" : "Cc1ccccc1",
"AMES" : {
"property" : "AMES",
"value" : -0.0042,
"PREDICTION" : "inactive",
"insideAD" : true
}
}
contains also information whether it is within the Applicability Domain of the model.