Back to Top

QSAR-models in USEtox

2 posts / 0 new
Last post
Biene's picture
QSAR-models in USEtox

The official USEtox User Manual represents that default QSAR-models in USEtox are applied to estimate input parameters. Comparing the results with parameters I found in databases they did not conform well. For instance BFAfish (benzene) = 4.266 L.kgfish-1 (database) and BFAfish (benzene) = 6.74 L.kgfish-1 (calculated) show differences. I do not feel comfortable to assess if these differences are considerable or not.

Is it all right to apply these QSAR-models to every substance even if it is a soluble or an insoluble one for example?

USEtox Team
USEtox Team's picture
Hierarchy of using and estimating USEtox input data

Thank you for your question.

For USEtox™ input data, we recommend to use data in order of reliability, where we recommend to use the following sequence:

  1. Experimental data (from literature studies, experiments, data bases)
  2. Specific QSARs, i.e. for specific substance groups such as brominated compounds, etc. (as specific as possible)
  3. Generic QSARs, e.g. for all organic compounds
  4. EPISuite estimation

If you have experimental data, always apply those! Only when calculating final characterization factors and performing an sensitivity analysis it can be known whether differences are considerable. Whether a QSARs can be applied in a specific case and is considered reliable has to be judged by expert judgement. Generally, uncertainty decreases from input data based on step 4 to 1 as described above.