-
Notifications
You must be signed in to change notification settings - Fork 66
Frequently Asked Questions
The main parameter to change is the temperature. You can also use the force button to add some of the non-clustered spikes to the already defined clusters. To recalculate using the GUI, delete the current times_filename.mat file.
The 'manual' button selects the spikes whose maximum or minimum over the selected samples resides inside the rectangle drawn manually. If the rectangle has a width of less than 3 samples, all the spikes with a sample inside the area will be select. Each "manual" button only works for the plot above and with its spikes.
There are few tricks to run faster. First, if you have too many spikes, you can do clustering on a first segment that will define the clusters, and assign the remaining spikes via template matching (the templates being the clusters defined in the first segment). This is set with the parameter max_spk. You can also use the option plot_average instead of plot_all, since plotting all spike shapes may take too long. If you just want to check a short segment of the file you can do this using the parameter tmax. If the functions of FilterM are in the path, Wave_clus will use them for reducing the time required for filtering the raw signal.
Sometimes Wave_clus chooses high temperatures that tend to overcluster the data. To avoid picking up high temperatures, you can set the parameter min_clus to a larger number.
Accept saves the clusters by assigning a corresponding integer (say 1 for the first cluster, etc.). Reject gives the cluster a value of 0; i.e. it will merge it with the non-clustered spikes.
The top number is the total number of spikes for the cluster, with the ones not assigned through template matching appearing in brackets. The bottom one is the number of spikes with an inter-spike interval (ISI) of less than 3ms. A relatively large number of spikes within 3ms ISI is a sign of a multi-unit cluster.
In this case as many extra figures as necessary will be generated.
Wave_clus is plotting only the first 2 wavelet coefficients. It may well happen that other coefficients will separate the clusters even better. You can actually plot 45 projections, corresponding to 10 wavelet coefficients, with the Plot all projections button.
Yes, you can edit, zoom, copy and paste, etc each plot using the Tools menu at the top of the GUI.
You can choose the parameters or load the default values. You should do that before loading the data.
Just decrease the segments_length for loading the data in smaller pieces.