Skip to content. | Skip to navigation

Sections
Personal tools
You are here: Home → Forum → Using data sets → a question on the single-unit spiking data in ac-3
 

a question on the single-unit spiking data in ac-3

Up to Using data sets

a question on the single-unit spiking data in ac-3

Posted by Fabiano Baroni at October 08. 2021
Dear authors,
 
I noticed that there are some very short ISIs, as short as 0.015 ms (so far I only looked at the first recording, site1-rn1.mat), which makes me wonder if the spike-sorting procedure has been validated. I haven't run an exhaustive analysis, but very short ISIs seem to be fairly common (for 27/31 single units in site1-rn1.mat , the shortest ISI is below 1 ms).
 
Thank you, best regards,

Re: a question on the single-unit spiking data in ac-3

Posted by Jermyn See at October 18. 2021

Dear Fabiano,

Thank you for your interest in the dataset. The spike sorting procedure used in this dataset is a semi-automatic spike sorter based on Lewicki (1994). Your characterization of very short ISIs in the dataset being "fairly common" is misleading, since while 27 of 31 single units have "short ISIs", only ~0.5% of all ISIs in that dataset is <1 ms. The manual part of the semi-automatic spike sorting procedure involves a curation of the isolated clusters from the automatic part of the spike sorter, and one of the criteria is indeed that single units cannot have a large percentage of short ISIs. We did not remove spikes that have short ISIs from the curated clusters, but please feel free to do so as you use the dataset.

Best,

Jermyn

Re: a question on the single-unit spiking data in ac-3

Posted by Fabiano Baroni at November 16. 2021
Dear Jermyn, thank you for the clarification. Best regards,
Fabiano

Re: a question on the single-unit spiking data in ac-3

Posted by Fabiano Baroni at November 18. 2021
Dear Jermyn,
 
I looked for very short (<1ms) ISIs across the whole shared dataset, and I could find some units with ~ 10% or more of ISIs < 1ms (that's the case for site7-rn1 , site3-rn1 also includes a neuron with similar values). I don't know where you set the threshold for removing single-units with a large percentage of very short ISIs, but 10% of ISI < 1ms seems quite concerning to me. But maybe the data can still be used, after removing some single units with many short ISIs and some spikes corresponding to very short ISIs from the remaining units.
 
After reading the crcns_ac-3_data_description.pdf and the corresponding ELife paper, it seems that the data available is only a small part of the data used in the paper. It only corresponds to ~ 10 min RN1 and RN16 stimulation experiments, while the paper also reports results from DMR experiments, as well as fifty 5-s repeats of RN or DMR, and spontaneous activity.
 
Could it be possible to obtain that data? Thank you, best regards
 
Fabiano
Powered by Ploneboard
Document Actions