Skip to content. | Skip to navigation

Sections
Personal tools
You are here: Home Forum Using data sets hc-3 - clu files - inconsistency in cluster counts (dataset ec016.233)

hc-3 - clu files - inconsistency in cluster counts (dataset ec016.233)

Up to Using data sets

hc-3 - clu files - inconsistency in cluster counts (dataset ec016.233)

Posted by Ruxandra Cojocaru at November 26. 2019

Hello.

 

From the documentation, I understand that the first line of a clu file is the number of unique clusters found in that file. I tried the UNIX commands given in the crcns-hc3-data-description.pdf page 5 and I could reproduce the behaviour for the files ec012ec.187.clu.1 and ec012ec.189.clu.1 used as an example.

However, for the dataset ec016.233, I found some inconsistencies.

For example:

ec016.233$ head -n 1 *".clu.1"; tail -n +2 *".clu.1" | sort -n | uniq -c
4
    606 0
 231543 1
    267 3

So there should be 4 clusters according to the first line, but only 3 unique clusters are present.

This also happens in clu.2, clu.3, clu.4, clu.9 and clu.10 of the same dataset.

Is there any reason/explanation for this?

 

Thank you,

Ruxandra

Re: hc-3 - clu files - inconsistency in cluster counts (dataset ec016.233)

Posted by Ze Henrique Targino at December 16. 2019
I am not the responsible for this data-set. But as far as I understood the spike sorting and cluster labeling is made over a day of recording.
 
Spike waveforms of all sessions within a day are stacked and the sorting is realized.
 
Some clusters ("neurons") may be present in a session but not in another one.
 
For this reason, the number of clusters may be the number of clusters within a day in that shank but not necessarily in that session.
 
Some neurons may be context dependent, or just the slight movement of the electrodes would lead to a change in the waveform and the emergence of a new cluster.

Re: hc-3 - clu files - inconsistency in cluster counts (dataset ec016.233)

Posted by Ruxandra Cojocaru at December 19. 2019

I understand and I believe you are right. I just was a bit confused, as the documentation rather says it's the number of clusters in the session specifically. Anyhow, I will just count them myself, as in this dataset the number of unique clusters in the header of the clu files really does not match the actual number.

 

Thank you for your reply!

Powered by Ploneboard
Document Actions