Skip to main content
QMRF y QPRF

We’ve Updated Our QPRF and QMRF Reports for Greater Reliability and Regulatory Compliance

As mentioned in previous blog post, quantitative structure-activity relationship (QSAR) models are essential computational tools for predicting the properties of chemical compounds. These models play a key role in transitioning to animal-free methods for assessing toxicity and the environmental impact of substances. QSAR studies allow us to meet the information requirements set by various laws and regulations. At our company, we apply QSAR models to regulations such as REACH, ICH’s M7 guideline y and the EU Ecolabel for lubricants.

Harmonization and Standardization in Regulatory Toxicology

One of the general principles in regulatory toxicology, particularly in today’s globalized world, is the harmonization and standardization of methods. This promotes the acceptance of results across organizations and countries, facilitates international trade, and reduces the need for experimentation. In this context, the Organization for Economic Cooperation and Development (OCDE) plays a vital role, as most parameters required for regulatory purposes must be tested following the guidelines set by this organization.

OECD Publications on Computational Methods

The OECD has published several documents on applying computational methods. In 2004, they defined the five principles a QSAR model must meet to be acceptable in regulatory evaluations. These principles were later developed in the OECD Guidance on the Validation of (Q)SAR Models. This guide also introduced specific formats for documenting models (QSAR Model Reporting Format, QMRF) and predictions (QSAR Prediction Reporting Format, QPRF), which were developed by the Joint Research Center (JRC) of the European Commission.

New Guidelines and the (Q)SAR Assessment Framework

This process has evolved, recently culminating in the publication of a new OECD guide in collaboration with experts from institutions such as ECHA and Italy’s ISS, titled the “(Q)SAR Assessment Framework: A Guide for Regulatory Evaluation of Structure-Activity Relationships (Quantitative), Predictions, and Results Based on Multiple Predictions (QAF)”. This new framework defines a checklist of evaluation elements to verify models and predictions, as well as a new format for QPRF reports.

ProtoPRED Updates

Our prediction platform, ProtoPRED, provides complete and detailed QMRF and QPRF reports to document predictions and meet the latest regulatory requirements. We’ve updated our reports based on the QAF, incorporating new sections with practical information to help evaluate the reliability of a prediction. This is a complex topic involving factors such as the overall predictability of the model, the target molecule’s relationship to the training set, the model’s performance with similar substances, and the model’s suitability for the specific application.

Additional Tools for Reliability Assessment

In addition to detailed explanations, the QPRF in ProtoPRED offers two additional tools to guide evaluation: a preliminary combined reliability score and a proposed evaluation of the QAF checklist. These tools are provided as guidance, based on pre-determined criteria, and aim to complement and support expert evaluation, which must consider external information and additional details described in the QPRF.

The reliability score in ProtoPRED is presented as a percentage, with a prediction labeled as “Reliable” if it exceeds 50% (or “Highly Reliable” if it surpasses 75%). The score is calculated as the average of a series of individual scores for relevant parameters. These include prediction reproducibility, overall fit quality, predictive performance, applicability domain, structural space, analog similarity, and local performance, among others.

If you’d like to learn more about our QSAR models and new QPRF reports, you can sign up for ProtoPRED and explore it freely with a trial credit for a limited time. Visit ProtoPRED to register.

Skip to content