Stability Updates

This update is the first part in a series of updates that are solely focused around stability and usability improvements.

As always, but especially now, we welcome any suggestions or bug reports. We are focused exclusively on stability improvements, usability improvements, and general user experience improvements for the next few months and would love to hear your suggestions.

Summary of Changes

• Progress reporting for most analysis steps
• Overhaul error messages
• Improve reliability of cluster heatmap
• Improve usability of scoring
• Allow exporting score tables

Progress Reporting

This update brings progress reporting to most steps of analysis:

1. DeepLabCut pose-estimation
2. Feature extraction
3. Skeleton video generation

A future update will provide progress reporting for the remaining step (preprocessing).

Error Message Overhaul

Gone are the days of full-screen errors when trying to generate graphs. This update provides brief descriptions for all error messages, and detailed messages in a separate dialog that can be accessed via the “View details” button.

We’ve also changed the aesthetics of error messages to make them less jarring.

Improved Reliability for Cluster Heatmap

We’ve further increased the stability of the cluster heatmap. This change builds upon the improved error messages by displaying small warning messages when certain aspects of the chosen summary CSV have data that are somehow incompatible with the procedure used to generate the graph.

There are a few ways a summary CSV might trigger warnings. Most of these are rare, but a few are more common. A non-exhaustive list of potential issues is:

1. Non-numerical values
2. Missing values
3. Constant values
4. Groups with one sample

Each of these cases will trigger a warning. To see which columns have been omitted, click the “View details” button on the warning.

Scoring Improvements

The Analyze > Scoring page has been updated to show which behaviors are tagged on the current frame. In the behavior list, you will see a purple dot next to each behavior which has been tagged in the current frame. This allows for exact control over which behaviors are tagged on any given frame.

Alongside this improvement, we’ve also implemented exporting the score table as a CSV file. When analyzing a capture, checking the “Data” checkbox in the “Export” section of the Analyze page will cause a new folder called scores to be produced which contains the score table for each scored capture.