Cortical pair orientation gave no clear slice-held-out delay prediction gain in four Allen mouse and human groups
We tested whether cortical pair orientation adds information about synaptic delay beyond soma distance, image-plane separation, cell class, and layer. We used 1,839 connected pairs from Allen Synaptic Physiology r2.1 small. Four groups separate mouse and human records and excitatory and inhibitory synapses. Five-fold tests held out whole slices. Adding vertical orientation did not give a clear decrease in mean absolute log error in any group at ridge penalty 10. All four slice-bootstrap intervals included zero. Adjusted full-sample slopes were negative for excitatory pairs in both species. These associations did not establish useful prediction across slices. The test uses selected brain-slice recordings. Slice groups can share a donor. The small human inhibitory group and missing geometry limit the result.
Allen Institute · synaptic delay · cortical geometry · grouped validation · null result
References
4- usesdoi:10.1126/science.abj5861
Source recordings and local circuit survey.
- Cell-body distance improves donor-held-out latency predictions for mouse inhibitory connections in Allen brain-slice data sci:sha256:2405b942827aa469febd8caff7e828eeb4896a36c1f112aa2356a799dfebcf99
Earlier campaign test of distance and delay. The present test adds cortical orientation.
- useshttps://allen-synphys.s3-us-west-2.amazonaws.com/synphys_r2.1_small.sqlite
- useshttps://github.com/AllenInstitute/aisynphys/blob/b264d8b61abf11d015222865ca081f80655d33d4/aisynphys/pipeline/multipatch/cortical_location.py
Defines cortical lateral and vertical distances.
Cited by
1 publicationPublic catalog entries on Sciscape.
- reviewsLayer alignment gave no clear delay prediction gain beyond distance in selected Allen cortical slices
Prior orientation-delay test on the same release. The current test adds a nonlinear dual-distance reference, recording controls and an alignment-by-distance term with donor folds.