In one V1DD mouse, shared inputs from imaged cells had no clear link to movie noise correlation in 6,416 pairs after controls

Shared inputs can make neurons vary together across trials. We tested this link in 6,416 pairs from one V1 Deep Dive mouse. The targets were 193 cells with proofread axons. The planned model controlled cell identity, distance, direct connections, and mean movie response similarity. A doubling of one plus the shared imaged-source count changed event noise correlation by 0.00051. The 95% dyadic interval was -0.00299 to 0.00400. It assumes independent errors between target pairs with no cell in common. Imaged sources supplied only 0.094% of recorded input synapses per target at the median. A later test used 905,533 incoming synapses and 675,197 source roots. Its all-source estimate was 0.00668. None of four source-panel tests passed Holm correction. Joint resampling of movie trials and time blocks gave ranges that included zero for every event source panel. Some fluorescence estimates remained positive. We rebuilt the movie measures from public traces and reproduced the source values within 3e-8. The result limits a sparse shared-input measure in this mouse. It does not show that common input has no effect. One mouse, source fragments, incomplete cell labels, and shared state limit inference. The package preserves compact graphs, trial responses, code, source hashes, figures, and repeat instructions.

connectome · shared input · noise correlation · mouse visual cortex · V1 Deep Dive · source coverage · trial variability · functional connectomics · reproducible benchmark