The main info
PASS Online predicts over 3500 kinds of biological activity, including pharmacological effects, mechanisms of action, toxic and adverse effects, interaction with metabolic enzymes and transporters, influence on gene expression, etc.
To obtain the predicted biological activity profile for your compound, only structural formula is necessary; thus, prediction is possible even for virtual structure designed in computer but not synthesized yet.
Prediction is based on the analysis of structure activity-relationships for more than 250,000 biologically active substances including drugs, drug-candidates, leads and toxic compounds.
Average accuracy of prediction estimated in leave-one-out cross-validation procedure (each compound is excluded from the training set and its activity predicted based on SAR model obtained on the rest part of the training set) for the whole PASS training set is about 95% (Filimonov and Poroikov, 2008).
Robustness of PASS algorithm has been shown in special experiments with principal compounds from MDDR database consisted of 18977 compounds with 124 activities. The set of compounds was 50 times divided at random into two equal subsets. The first subset was used as the training set, the second one as the evaluation subset and vice versa (100 experiments). 20, 40, 60, 80% of information (activity/structure data) were randomly excluded from the training set. Average accuracy of prediction (IAP) was calculated for each type of activity. It was shown that, despite the removal of up to 60% of information, PASS still provides a reasonable accuracy of predictions (Poroikov et al., 2000).
Since PASS service is used by medicinal chemists, pharmacologists and toxicologists for several years (Lagunin et al., 2000), there are many publications where PASS predictions were confirmed by subsequent synthesis and biological testing.
To provide more accurate predictions for compounds belonging to new chemical classes and to extend the predictable area onto new biological activities, we are permanently working on enlargement of PASS training set.
If you want to improve the quality of PASS predictions for compounds from your chemical series and/or your biological activity, you can add information about relevant biologically active compounds, which you know, into the PASS training set. The next PASS version will include knowledge about structure-activity relationships extracted from your information.