Due to the way how the system is programmed and set up the data remains stored on the web server, but is not freely accessible or browsable by third parties. Each calculation is assigned unique and random URL which is only known to the user who submitted the calculation. Server administrators have full access to all submitted data but it will be only used for debugging purposes. Users concerned with data safety can run the tool locally on their computers since all source code and models are available on GitHub.
This site creates csrftoken cookie. This is an essential cookie used to protect against cross site request forgery and cross site scripting.
Endocrine Disruptome is an open source, user friendly and web based prediction tool. It uses molecular docking to predict binding of compounds to 14 different nuclear receptors: androgen receptor, estrogen receptors α and β, glucocorticoid receptor, liver X receptors α and β, mineralocorticoid receptor, peroxisome proliferator activated receptor α, β/δ and γ, progesterone receptor, retinoid X receptor α, thyroid receptor α and β.
Endocrine Disruptome runs on the platform called Docking interface for Target Systems (DoTS). It uses AutoDock Vina in the background to perform dockings. Dockings are performed in 18 previously validated structures.
Click on prediction and either draw a structure or submit SMILES. Upon submission of the structure the you are immediately redirected to a docking page assigned to your calculation. Instantly you will see the structure, calculated molecular properties (such as molecular weight, logP, hydrogen bond acceptors and donors) and PAINS alerts. While docking is running in the background (you see the message in blue box), the page will automatically be refreshed every 30 s. Usually docking should be finished in 5 - 10 minutes, depending on the complexity of the molecule. The color coded table will be displayed after the docking.
The 18 structures integrated in Endocrine Disruptome are carefully selected from 103 crystal structures. All structures and protocols are validated. Users are advised to consult the receptors' sub-pages to check the performance of each model. The most important parameter is area under the curve (AUC) under receiver operating characteristic (ROC) curve. Models with higher AUC values should perform better. Results are color coded and binned in 4 classes.
Endocrine Disruptome is intended to be used for screening. Therefore classes were divided in a way to more correctly predict negative results, i.e. compounds that don’t bind to receptors.
Three thresholds were set per structure to enable a division to 4 probability binding classes: red, orange, yellow and green. For the threshold calculations sensitivity (SE) was used to obtain corresponding docking scores. Class “red” corresponds to SE < 0.25 and indicates high probability of binding, two intermediate classes “orange” (0.25 < SE < 0.50) and “yellow” (0.50 < SE < 0.75) indicate medium probability of binding, and class “green” (SE > 0.75) corresponds to low probability of binding.
For an example of bisphenol A docking results please see here.
The article about Endocrine Disruptome and DoTS was accepted in JCIM:
KOLŠEK, Katra, MAVRI, Janez, SOLLNER DOLENC, Marija, GOBEC, Stanislav, TURK, Samo Endocrine disruptome - an open source prediction tool for assessing endocrine disruption potential through nuclear receptor binding. Journal of chemical information and modeling, 2014, vol. 54, iss. 4, p. 1254-1267.