ProtoPRED

July 2024

New models:

We welcome a new models to the ProtoPRED server: Acute oral toxicity (2000 mg/kg bw).

ProtoREACH incorporates a new model for Acute oral toxicity and now it has two different classification models. The models are based in the same endpoint but classifiy the substance with a different threshold. The new model has a threshold of 2000 mg/kg bw, which corresponds to the limit between classification as toxic (categories 1-4) or non classified in CLP. On the other hand, the preexisting model, which has been renamed to Acute Oral Toxicity (300 mg/kg bw), has a threshold of 300 mg/kg bw and can be used to discern between medium/high toxics (categories 1-3) and low or no toxicity (category 4 or not classified).

Reporting of results:

We have updated the reports provided with your prediction. New QPRF reports, based on the 2023 (Q)SAR Assessment Framework of the OECD are provided. New QPRF files are more complete and include additional information about your prediction to help the user and external evaluators to assess the prediction. Among other upgrades, those reports include a new reliability score to provide an orientative guide on the quality of the calculation and a preliminary assessment related to the OECD prediction checklist.

More:

We have improved the management of user data. Now, the user can change its password in the user section and recover it from its email (note that this is the registration email, not the one provided for QPRF reports).

We have improved the privacy of your data. The historic of the predictions done by an user is saved and offered to the user through the user page as a feature to keep track of usage. By default this includes the SMILES and predicted value, to facilitate revision. However, the user can select the Data privacy option to stop saving their results (the date and model will be stored, but not the substance or predicted results).

Performance and appareance has been improved by fixing other minor bugs and enhancing the design. Minor typos have been corrected.

November 2023

New models:

We welcome a new module to the ProtoPRED server: GenoITS.

The GenoITS module presents a computational workflow focused on the prediction of genotoxicity using the Integrated Testing Strategy proposed by REACH. The ITS provides a guide about how to combine the tests in five different categories (gene Mutation in bacteria; in vitro and in vivo and cytogenicity in vitro and in vivo) and their interpretation. Hence, GenoITS uses up to 5 different QSAR models and/or external data provided by the user to follow the complete workflow and provide a single prediction for genotoxicity.

Reporting of results:

The QPRFs have been updated to include additional information such as a summary table with the most relevant data: Endpoint, Target molecule, Predcition and Applicability Domain assessment.

A direct link to the QMRF of each model is presented with the results tables, so the details on the endpoint and the database are also accessible after prediction.

More:

Performance and appareance has been improved by fixing minor bugs and enhancing the design. Minor typos have been corrected.

July 2023

New models:

We welcome a new module to the ProtoPRED server: ProtoNANO.

The ProtoNANO module includes nanoQSAR models, i.e. QSAR models adapted to their use in inorganic nanomaterials. Several endpoints are already available including cytotoxicity of metal oxides in bacteria, cytotoxicity of quantum dots in healthy and tumoral human cells and relevant physicochemical properties such as the partition coefficient and the Zeta potential. Note that due to the particularities of nanomaterials, nanoQSAR models required additional input, such as experimental values for the size.

Reporting of results:

We have updated the information provided with your prediction. A summary table has been added to check multiple-endpoint predictions in a single table. Also, additional columns have been included both in the web and excel version for some regression models regression models, such as commonly used units (if different than the model unit) and the value without units (to facilitate automatization of the analysis).

More:

Performance and appareance has been improved by fixing minor bugs and enhancing the design. Minor typos have been corrected.