Additional steps-2, Step 7

Additional steps:
II) Searching a dataset according to prediction values

You can search for specific prediction data in Ambit2 by using the following URL construct:

Simply replace with the feature number within which you'd like to search, and the with what you want to search for.

Let's search for all Class I predictions obtained with the ToxTree: Cramer rules model. The feature number is 22254 (the feature URI is, and the search query is Low+(Class+I). Note that the "+" sign stands for a space. The resulting search URL is:

The Cramer rules for oral toxicity are very conservative, and negative predictions (i.e. as "Low (Class I)") can be used with fairly high confidence. Thus, this small set represents a good set of potentially safe compounds to study further. However, keep in mind that even though a compound may be classified as "High (Class III)" by Cramer rules there is a good chance it is wrong. Summarizing, having a compound classified as "Low (Class I)" is quite relevant information, whereas having a compound classified as "High (Class III)" is not.

Alternatively, if we are interested in eliminating toxic compounds from our drug candidate list, we could search for compounds classified as Class I (narcosis or baseline toxicity) according to the Verhaar scheme. The feature number for the "ToxTree: Verhaar scheme for predicting toxicity mode of action" model is 823450, and the search query is Class+1+(narcosis+or+baseline+toxicity). The resulting URL is

Note that these search links display the filtered dataset according to the search query, but they don't display the prediction features according to which the search has been made. To include those, append &feature_uris[]= (for the Cramer rules) or &feature_uris[]= to the search URLs.


Additional steps III) Obtaining additional predictions

Step 1: Creation of an OpenTox dataset from a local file
Step 2: Selection of models available in OpenTox through ToxPredict
Step 3: Application of the selected models to the uploaded dataset
Step 4: Obtaining simple statistics for the predictions obtained for the uploaded dataset
Step 5: Adding the predicted values to a tabular view of the dataset and downloading the table
Step 6: Additional steps I) Applying an individual model to a dataset
Step 7: Additional steps II) Searching a dataset according to prediction values
Step 8: Additional steps III) Obtaining additional predictions
Step 9: Additional steps IV) Downloading data from other dataset services in OpenTox
Step 10: Complement your findings with predicted associations with hepatobiliary adverse events


BACK TO Build a Weight of Evidence for Drug Candidate Molecules