Session 7: Assessing Prediction Reliability

Applicability Domain Analysis (ADAN): A Robust Method for Assessing the Reliability of Drug Property Predictions
PRESENTING AUTHOR: 

Pau Carrió

INSTITUTION / COMPANY : 

FIMIM

POSITION: 

PhD Candidate

ABSTRACT CONTENT / DETAILS: 

There is a pushing need in the pharmaceutical industry to develop methods able to assess the safety of drug candidates at early stages of development.

In silico approaches are promising, but applied to toxicity endpoints do not perform always as expected.

The eTOX project - a public-private partnership project within the IMI framework- aims to address those needs by developing a drug safety database from the pharmaceutical industry legacy toxicology reports and public toxicology data; innovative in silico strategies and novel software tools to better predict the toxicological profiles in early stages of the drug development pipeline.

Our work in the eTOX project has focused to improve currently used in silico prediction methods for their application to biological endpoints, with a special emphasis to toxicological endpoints.

Here, we report a novel general methodology called ADAN (Applicability Domain Analysis) for assessing the reliability of drug property predictions obtained by in silico methods.

The assessment is based on the comparison of the query compound with the training set, using six diverse similarity criteria and allows classifying the query compound within reliability categories.

The validation results - performed in unfavourable conditions, like the prediction of drug toxicity endpoints- confirm the robustness of the proposed methodology.

Furthermore, we propose a unifying strategy for the use of in silico predictive methods applied to toxicity data sets.