search

cortex-lab / KiloSort

GPU code for spike sorting
GNU General Public License v2.0
172 stars 100 forks source link

Redefining principal components #32

Open SheinIdelsonMark opened 7 years ago

SheinIdelsonMark commented 7 years ago

Hi

I was wondering if you have a script for calculating the predefined PCs. You mentioned in the config file that these should be recomputed for new data sets. Is this just the weight matrix for a sample of representative spikes? Are the PCs computed on raw data or filtered data (which filter)?

Many many thanks, Mark

marius10p commented 7 years ago

Hi Mark,

Yes, these are just the PCs of a sample of representative spikes (just their largest channels), aligned at sample 20. They are the temporal PCs, not the spatial masks. You can see it's just a .mat file that is loaded from disk.

I should really add the PC computation to the preprocessing: will do so at the next update. It really should not play much of a role in the sorting itself, but it might make the PC visualizations in Phy clearer.

Best, Marius

SheinIdelsonMark commented 7 years ago

Hi Marius,

Thanks for the info. If this will not create major issues in Phy, I will just wait for your next update.

Many thanks and all the best, Mark

On Mon, Oct 24, 2016 at 11:03 PM, Marius Pachitariu < notifications@github.com> wrote:

Hi Mark,

Yes, these are just the PCs of a sample of representative spikes (just their largest channels), aligned at sample 20. They are the temporal PCs, not the spatial masks. You can see it's just a .mat file that is loaded from disk.

I should really add the PC computation to the preprocessing: will do so at the next update. It really should not play much of a role in the sorting itself, but it might make the PC visualizations in Phy clearer.

Best, Marius

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/cortex-lab/KiloSort/issues/32#issuecomment-255865297, or mute the thread https://github.com/notifications/unsubscribe-auth/AV939flrbbh2BC7n2T46Smee0x80WHgXks5q3R0FgaJpZM4Ke4V9 .

Mark Shein-Idelson (PhD) Max-Planck-Institute for Brain Research Department of Neural Systems & Coding Max-von-Laue-Str. 4 60438 Frankfurt am Main Germany

email: Mark.Shein-Idelson@brain.mpg.de institute homepage: www.brain.mpg.de personal homepage: https://sites.google.com/site/marksheinidelsonwebsite/ phone: +49 - 69 - 85 00 33 - 2030