-
Notifications
You must be signed in to change notification settings - Fork 222
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Hard-coded spacing constraints crashing Kilosort 3.0 #313
Comments
That was supposed to be dmin indeed., I fixed it now. I also added a horizontal pitch detection for spike templates in the spike extractor, so that in your case it should use three different X positions: 0, 25 and 50. |
Please let me know if the latest commits have fixed this. |
Not quite working. Currently referencing ...replacing with simple line of code from extract_spikes.m: dNearActiveSite = median(diff(unique(rez.yc))); did the trick, but wouldn't we actually want something that also takes x distance into consideration? ...as I'd expect Not sure I follow the "three different X positions" statement from your previous comment. xc = repmat( [-25;25], nChanTOT/2, 1); ...is that not correct/compatible? |
Thanks, fixed it now. That condition will only kick in for non-grid-like geometries, so it should be fine in most cases to set it to a reasonable number like the vertical pitch. "ycup, xcup" are the centers of the template prototypes used to extract spikes both for drift correction and for the first clustering. This spike detector is described in the Neuropixels 2 paper. It's good to have some that are centered in-between sites, but the defaults were only good for Neuropixels. Now they should be ok for you as well, with templates centered at half-integer site spacings horizontally and vertically. Also, I forgot to make them consistent between datashift2 and extract_spikes, and I updated that just now. |
Hard-coded spacing constraints crashing Kilosort 3.0
... seems to be the issue, but please confirm & consider
Running Kilosort 3.0 on 32ch U-probes (stereotrode config: 50um within [x], 100um between [y]; 1,500um total length)
Crashing during find_merges.m
rez.st3(:,2)
contains many zeros that originate during final_clustering.mst3(:,2)
is replaced byhid
, which is derivative of run_pursuit.m outputs< 20
Presumably this should/could be something like
dmin
instead...The text was updated successfully, but these errors were encountered: