In order to facilitate considerations on the adequacy of the prediction (model result) in relation to a defined regulatory purpose, JRC has compiled a standard for reporting (Q)SAR predictions for chemical compounds.
The QSAR Prediction Reporting Format (QPRF) is a harmonized template for summarizing and reporting substance-specific predictions generated by (Q)SAR models.
Q-edit is a guided user interface that couples the various OpenTox web services with the QPRF standard thus expediting the work of creating proper QPRF reports.
The QSAR Prediction Reporting Format (QPRF) is a harmonised template for summarizing and reporting substance-specific predictions generated by (Q)SAR models (check the specification here). Q-edit is tool that facilitates the whole procedure of prediction generation and reporting combined with database lookups based on the OpenTox web services.
The OpenTox report on "Automated Validation Report Generation" available here contains the background and progress which has been achieved within OpenTox with respect to the creation of automated reporting facilities for validated (Q)SAR models of predictive toxicology data.
In this article, we will start by reviewing the key requirements of a QPRF report and then proceed with the necessary steps to generate a QPRF report using Q-edit - a new QPRF editor developed under OpenTox to facilitate the creation of QPRF reports by an end user.
Users can download the latest version of the OpenTox QPRF editor (Q-edit) from here. Developers can access the source code at the CVS https://github.com/alphaville/Q-edit. Use the https git directory https://github.com/alphaville/Q-edit.git or the git read-only version: git://github.com/alphaville/Q-edit.git.
Users have to have installed some Java Runtime Environment and Virtual Machine on their operating systems. Until now we have tested Q-edit with Sun Java version 184.108.40.206 and 220.127.116.11. Normally users should not encounter any problems trying to run Q-edit on any other Sun JVMs of newer version or other JVMs.
Users that need to access protected resource in OpenTox (read the article posted at http://opentox.ntua.gr/index.php/blog/5-authentication on Access Control management in OpenTox) should install the SSL certificate for opensso.in-silico.ch on their machines. You will definitely find the information provided here useful.
You will find the latest version of Q-edit at.http://opentox.ntua.gr/files/Q-edit.jar. Download it and run it locally.
On Linux machines using the terminal, use the command
java -jar ./Q-edit.jar. It is possible that Linux users (and in particular on Ubuntu as it has been observed) need first to modify the permissions of the downloaded file to render it executable. In this case, the following warning message will pop up:
In this case right click on the Q-edit.jar file and choose "Properties", once opened go to the "Permissions" tab and check "allow executing file as program":
Linux users may use the following commands to download and run Q-edit :
wget http://opentox.ntua.gr/files/Q-edit.jar chmod +x ./Q-edit.jar java -jar Q-edit.jar
Using Q-edit, users will be able to:
Once you launch Q-edit, create a new report (File > New Project > Empty Report or just Ctrl+N). You should be able to see the new report in the desktop of Q-edit:
Now enter the identifier of a compound in the text field as in the following figure:
Strike the return key or click the Load button to retrieve all available information from the online OpenTox web services at http://apps.ideaconsult.net:8080/ambit2. The details are loaded on the same view and are presented as follows:
The "Details" button gives you access to the various structural attributes of the compound like its SMILES string as well as to other identifiers like the CAS registration number and the INECS number as show in the next figure:
The user is then expected to discuss on the stereochemical structural attributes of the compound that can possible affect the reliability of the prediction. The "Stereo" button in the toolbar of the same view will open a new dialog box in which this information should be provided:
A list of synonyms is also loaded but the user might need to add one or remove some synonym that is invalid according to their opinion. For example in this example, the synonym
Molecule-1 needs to be removed:
You can add a synonym of your own using the green (+) button:
Finally, note that one can look up for a compound using directly its OpenTox URI, for example try http://apps.ideaconsult.net:8080/ambit2/compound/24793/conformer/372927 instead of "carbasalate calcium".
The authors of the reports are included using the wizard you'll find under the second tab:
Fill in your name, email and other information as requested and click OK. The first two fields are mandatory. Also note that you cannot add the same author (i.e. an author with the same name) more than once. Select any author from the list on the left to see their data:
You can also update/edit the information you previously submitted for an author using the "Edit Author" button:
Set the date when the report was created before proceeding to the next step.
If you have a URI of an OpenTox model resource, you may enter it directly to the corresponding field in Q-edit:
OpenTox services list the URIs of all available models under /model. For instance check out http://opentox.ntua.gr:8080/model and http://apps.ideaconsult.net:8080/ambit2/model. You can also search for approved models that are also included in ToxPredict at http://apps.ideaconsult.net:8080/ontology. Copy a URI and paste in Q-edit as shown in the figure above. Then click the "Load" button or strike enter to load all information available for it. In case the model you chose is password-protected you will need to provide your credentials. The following dialog box will appear:
In case you don't have an account, you may just "Login as Guest" clicking on the corresponding button. Afterwards, you can inspect details about the loaded model - just click on the "Details" button that appears in the Model tab toolbar:
Therein one may find the Model's title (e.g. here it is "MLR model for LC50_mmol"), the prediction, dependent and independent features of the model, a link to its training dataset, all identified by their URIs. Later we'll show how to access more human-readable information about all these. If the model has a certain parametrization, this can be found in the second tab, namely "Model parameters":
Information regarding the training algorithm that was used to create the model can be retrieved by clicking on the "Algorithm" Button:
and the same way one can inspect all available information for the predicted and dependent features of the model:
You may use Q-edit to find models that reside (are provided by) in various on-line web services or are hosted by the central ontology server of OpenTox. For that click the button reading "Find Online" under the "Model, Model Info" tab:
This will fire up a dialog box that allows you to choose your model from a list. Click "Go!" and you'll get a list like the one illustrated here:
Using the compound and the QSAR model provided by the user in the previous sections, Q-edit can invoke the necessary OpenTox web services to return the predicted value as well as (if available) the experimental value for the compound. This functionality is available under the fourth tab - namely Prediction - where the user is required to report the predicted value and comment on the uncertainty of the prediction and the possible chemical and/or biological mechanisms that might affect its reliability - for example a particular compound might appear to be active according to a QSAR model but in practice it is not well absorbed.
A list of structural analogues can be retrieved on the basis of some similarity index provided by the user. Under the tab "Applicability" find the tab "Structural Analogues" where one can provide a similarity threshold in the range 0.5 - 1.0 and click on the button "Acquire List" to get a list of compounds that are similar to the one submitted in the beginning. This is illustrated in the following screenshot where the analogues of sucrose (up to 95% similarity) are listed.
If a model is loaded, then one can download all experimental values for each one of the structural analogues. Finally, after the list of structural analogues is loaded and one of them is selected, one can inspect its structural information available by clicking on the button "Compound Info" which is found in the toolbar.
We're pretty fast in fixing issues with Q-edit, so if you notice any bugs or have some improvement in mind please file an issue at https://github.com/alphaville/Q-edit/issues - for that you need a github account which you can create on github. Github will automatically notify you by email once there is an update on your issue.