-
Notifications
You must be signed in to change notification settings - Fork 18
GUI Keys
One of the main advantages of Combinato is that it has a relatively efficient GUI that enables you to modify sorting results, and displays a variety of visual cluster quality checks. These plots are updated on the fly, so that you can instantly see how a change in clustering influences various properties of the clustering.
Start the GUI by calling css-gui. Make sure to start the GUI in a folder containing the folders that contain the data_xyz.h5 files. For example, if you have a file /home/myname/experiment1/channel5/data_channel5.h5, first move to experiment1, then start css-gui.
Use one of the following commands to load a sorting session. You can either open a sorting session with File-Open or you can use job files.
Here, a job file is simply a list of paths to data_xyz.h5 files, saved as do_manual_pos.txt or do_manual_neg.txt, depending on the polarity of spikes. If you use job files, you can move from one sorting session to the next by key press, which can speed up the sorting review process considerably if you have many recording channels.
| Menu | Command | Shortcut | Description |
|---|---|---|---|
| File | Open | Ctrl+O | Load a sorted session for optimization |
| File | Open Jobs | Load a list of sorting jobs | |
| Actions | Next Job | Ctrl+J | Load next job from joblist |
| Actions | Goto Job | Load a specific job from joblist | |
| File | Save | Ctrl+S | Save changes to file |
Once a sorting session is loaded, the various groups of clusters appear as a pull-down menu. If you select a group in the pull-down menu, information about it is displayed in the different tabs.
This tab collects informative plots about one group of clusters. Some of these are inspired by [this article][8] by D. Hill, S. Mehta, and D. Kleinfeld.
Press Enter to additionally highlight information about a single cluster within a group. The information in this tab helps decide if a group (or a cluster in a group) is an artifact, or a multi or single-unit. Use the information in the following way.
| Subplot | How to use |
|---|---|
| (log) Density | If the density is inhomogeneous across sampling points, it is more likely that the group is a multi-unit. |
| Mean spikes | If some of the mean spikes differ a lot in shape, consider splitting the group into two or more groups. |
| Inter-spike intervals | If you see a clear comb-like pattern with peaks at 20, 40, 60 ms, the unit is an artifact or contains clusters that are artifacts. If the percentage of lags < 3 ms is high (> 5 %), it is unlikely that the group is a single-unit. |
| cumulative spike count | If the cumulative spike count contains clear |
This tab displays all groups, by plotting the mean spikes of all clusters in a group to represent that group. The main function of this tab
| Menu | Command | Shortcut | Description |
|---|---|---|---|
| Actions | Merge all | Ctrl+M | Merge all clusters into group 1 |
| Actions | New Group | Ctrl+N | Create a new group, to which clusters can be added |
| Enter | Highlight information about selected cluster | ||
| N | Move to next group in group list | ||
| A | Move selected cluster to artifacts group | ||
| G | Move selected cluster to closest group |