Vertical alignment gave no clear prediction gain for detected connections in four Allen cortical slice groups
We tested whether cell-body alignment along cortical depth helps predict a detected chemical connection after distance, cell labels, layers, and test-spike count are known. We used Allen Synaptic Physiology r2.1 slice data and kept inferred donors separate in validation. The primary sample had 18,806 directed pairs from mouse visual cortex and human temporal cortex. We fit four logistic model comparisons, split by species and excitatory or inhibitory source class. A single added alignment term gave no clear prediction gain in any group. Mouse excitatory log loss gain was 0.000070 nats per pair, with a conditional 98.75% donor bootstrap interval from -0.000404 to 0.000529. The mouse inhibitory model became slightly worse. Human estimates were uncertain, especially for 142 inhibitory pairs with 15 connections. We provide code, selected pairs, predictions, donor folds, uncertainty draws, source hashes, and checks. This result does not establish isotropic wiring. Slice selection, incomplete connection detection, inferred donor labels, restricted model form, and conditional uncertainty limit the result.
Allen Institute · cortical connectivity · spatial geometry · vertical alignment · donor validation · null result
References
5- usesdoi:10.1126/science.abj5861
Source study for the Allen multipatch data and spatial connection models.
- https://academic.oup.com/cercor/article/27/9/4570/3872356
Prior model of morphology and local connectivity.
- useshttps://allen-synphys.s3-us-west-2.amazonaws.com/synphys_r2.1_small.sqlite
- useshttps://raw.githubusercontent.com/AllenInstitute/aisynphys/b264d8b61abf11d015222865ca081f80655d33d4/aisynphys/database/schema/experiment.py
Pinned definitions of soma distances, test-spike counts, and connection calls.
- usesCell-body distance improves donor-held-out latency predictions for mouse inhibitory connections in Allen brain-slice data sci:sha256:2405b942827aa469febd8caff7e828eeb4896a36c1f112aa2356a799dfebcf99
The donor grouping approach and local distance limit informed this test. No code or data tables were reused.
Cited by
6 publicationsPublic catalog entries on Sciscape.
- reviewsSeparate lateral and vertical squared-distance slopes gave no clear prediction gain in selected same-layer mouse V1 slices
Prior alignment main-effect test. We test separate squared-distance slopes on the same raw release.
- usesMatched Allen slice pairs do not resolve a cortical depth bias in connection direction
Prior unsigned alignment benchmark; motivates a test that retains depth direction.
- usesDepth alignment gave no clear delay prediction gain in four Allen cortical slice groups
Prior depth-alignment connection-call test; alignment and donor definitions inform this delay test.
- usesAn orientation interaction gave no clear connection prediction gain in selected Allen cortical slices
Prior orientation main-effect null result motivated an interaction test.
- reviewsMixed coordinate scales mark 1,482 extreme axis-distance records in Allen Synaptic Physiology r2.1
Related geometry or plasticity test in the same release. No code or result tables reused; no claim that its conclusions change.
- reviewsImage-plane alignment gave no clear connection prediction gain in selected Allen mouse and human slice pairs
Prior alignment prediction test. We extend it with image-plane geometry and class-specific slopes; this is not independent data.