Including contacts from the same source changed the sign of an orientation similarity score on 354 MICrONS targets with tuning assigned by source

We tested whether close synapses from different source neurons have similar visual orientation estimates. We scanned all 337,312,429 records in the MICrONS v1507 filtered synapse table. We kept 328,340 contacts from 3,437 selected sources. The planned test used 354 targets with cleaned dendrite labels and both close and reference source pairs. Close meant at most 5 micrometres. Reference meant greater than 20 and at most 50 micrometres. Each distinct source pair entered once per target, using its minimum contact distance. The mean orientation similarity contrast was -0.06932. The exact mean under source record shuffles within area, cell type and 100 micrometre soma cubes was +0.00801. The upper-tail probability was 0.972 from 999 draws. A planned broader target check and a readout-center test also gave no clear excess similarity. An added description used individual contact pairs on the same 354 targets. Including pairs from one source changed the orientation contrast from -0.07080 to +0.12872. With equal target weights, one-source pairs formed 22.30% of close pairs and 1.89% of reference pairs. Their similarity was 1 by definition under a source tuning assignment. This positive contact score alone cannot establish grouping of distinct inputs. The results use model estimates and straight-line distances in one mouse. They do not establish a lack of functional grouping on dendritic branches. The package preserves the compact location data, code, source hashes, all draws, figures and repeat steps. A full raw-data scan matched every selected contact field. A replay reproduced 29 numerical result files exactly.

connectome · MICrONS · visual cortex · synapse location · orientation tuning · source identity · conditional permutation · reproducible dataset