Loading...
Loading...
Contributors:
Dataset URI:
10.5281/zenodo.23375856This dataset accompanies a study of overlapping neuronal-avalanche dynamics in layer 2/3 of awake mouse visual cortex. Mesoscopic two-photon calcium imaging was used to record neuronal populations across striate and extrastriate visual areas during spontaneous activity and visual stimulation. Convolutional non-negative matrix factorization (seqNMF) separated co-occurring spatiotemporal activity patterns, whose avalanche dynamics were related to pupil-linked arousal and measures of proximity to criticality and information processing. The deposit contains merged (2 paths from diesel2p) fluorescence-change (Delta F/F) matrices and deconvolved spike-rate estimates for the 10 analyzed recordings. Nineteen matrices are distributed as float32 arrays. The tigre806 spontaneous-ball spike-rate matrix retains its original float64 precision because float32 rounding changed one decision at the published detection threshold. Cell masks, coordinates, timestamps and available pupil/stimulus information retain their supplied precision. The repository also contains selected seqNMF factorization fits, sequence-level metrics and compact original-precision plotting inputs, together with Python figure-rendering code, archived analysis notebooks, MATLAB seqNMF code and environment information. The selected MATLAB fits preserve the actual factorization results used in the analysis. The documented fitting workflow uses K = 20 candidate factors, a motif length of 10 frames, 100 iterations, and a sweep over 100 logarithmically spaced regularization values from 0.0001 to 0.01. The saved reconstruction workflow selects a regularization value by balancing normalized reconstruction and regularization costs. Historical input preparation and sequence reconstruction included float16 quantization; this is documented rather than silently altered. The release conversion to float32 applies only to the Delta F/F and spike-rate matrices. Large sequence-reconstruction matrices and redundant intermediate caches are omitted. The small selected fits and plotting inputs are included so that the archived figures do not depend on obtaining identical factors from a new random initialization. The figure-rendering workflow is available in Python; refitting seqNMF requires MATLAB and the supplied seqNMF implementation.
Keyword(s):
MeSH Term(s):
Grants and Funding:
No Grants or Funding Data Available
Associated Publication(s):
Supplementary Data:
No Supplementary Data Available
Related Dataset(s):
No Related Dataset Data Available
The National Library of Medicine (NLM) provides information about biomedical datasets. Inclusion on an NLM Web page does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: NLM Web Policies (nih.gov)