Please log actions of user

[jdeligt]

In the light of reproducible research it would be good to log the actions performed through the GUI. Having to keep track of all settings/action will require a lot of documentation/logging by the user to be able to use the obtained results in their research.

[urmi-21]

Hi,
ORIS doesn't log the user actions performed via GUI. The user is expected to log information on their own when they find useful results (e.g., exporting the location of origin using auto-correlation method) and the results could be easily reproduced if the user provides the data, the methods and the parameters used.

ORIS is developed for interactive exploration of whole-genome sequences and it provides specialized methods for finding the origin of replication sequences. This process of exploratory data analysis is usually non-linear and may take several steps until a conclusion is reached. This can generate very big logs and most of the details in the log may be of little importance. However, the results which matter to the user could be easily exported when required. We leave it to the user to best describe/document the results and these results may be easily reproduced again using the same data and the methods.

Logging all the user actions may enable some advance users to keep track of the actions performed but I think ORIS’ ability to export results along with user documentation is sufficient for reproducibility.

If one wants to rely on logs then again users need to take an extra step to mine the useful parts in the log file. I agree that this could be complementary to what ORIS can do right now but not having this doesn’t limit the applicability of the software.

I would also like to say that the GUI itself is designed so that the software is easy-to-use for the people who are not familiar with coding.

The settings, are saved in a file called settings.cfg which is located in the same directory as the .jar file.

[jdeligt]

Thank you for the response and explaining the rationale. And just to be clear I think it is great that you went throught the effort of making a GUI to enable people to do these types of analysis in a more intuative and interactive way.

However many interactive/online tools do log user actions and some can even reconstruct the path to a figure even when there is al lot og back and forth in settings/algorithms involved. Yes this is quite some work but it does allow the user to download a log of commands/settings that resulted in their fiigures/conclusions. This is great for publishing and in some cases allows re-using of certain steps without all the clicks.

It might be a matter of perspective/opnion but I would argue that these logs are especially relevant to individuals not familiar with coding. They might not be as aware of the impact the choice of profile search/algorithm might have on their conclusions. It's our resposibility as developers to make them aware of these and provide them with ways of seeing what is happening.

Please understand that this is not a ctitique on the usefullness of your GUI but to a common trend in interactive anlysis. These types of analysis are powerfull because the put tools in the hands of the people with the biological questions but they suffer from reproducibilty issues.

I'll leave the issue open so that users can be aware of the fact that they will have to do their own regious logging and decide for themselves wether this is an issue for them or not.