EASI-PASS: An accessible pipeline for linking functional imaging and mRNA profiling

Key results

1. 01

   ShownEASI-PASS found a matching EASI-FISH cell for most functionally imaged neurons in both preparations, giving thousands of matched cells per experiment.

   In vivo visual cortex: 22,421 of 27,152 two-photon cells matched (82.6%; n = 5 mice). Acute parabrachial slices: 14,728 of 20,397 cells matched (72.2%; 7 slabs from 7 mice). Each in vivo experiment yielded ∼150-250 simultaneously recorded interneurons per class plus ∼4,000 putative excitatory neurons. · Abstract; Cell matching across 2P and EASI-FISH; Distinct arousal tuning and spatial correlations…; Figure legends

2. 02

   ShownAfter manual landmarking, stepwise refinement by mask intersection over union (IoU) improved alignment in every sample. Soma-print, which matches cells by the geometry of their neighbours, recovered more cells than a fixed IoU threshold and mostly picked the same partners.

   Compared with the landmark-only baseline, the final alignment raised median IoU (paired t test, slice p = 1×10⁻⁵, in vivo p = 0.002), the Soma-print matched fraction (p = 0.001; p = 0.002) and the AUC separating best from second-best scores (p = 0.03; p = 0.008). Soma-print and IoU > 0.3 chose the same partner in ∼94% of cells both matched; Soma-print matched ∼9% more cells. · Cross-modal registration approach; Cell matching across 2P and EASI-FISH; Figure legends

3. 03

   ShownIn mouse parabrachial nucleus (PBN) slices, neurons excited by optogenetic stimulation of lumbar spinoparabrachial axons were enriched for Syt10. Syt10 stayed the best predictor of the evoked response after accounting for each neuron's position and local axon density.

   1,242 of 9,024 matched PBN neurons (13.7%; 6 mice) were activated. Syt10 (p = 0.006) and Th (p = 0.007) were higher in activated neurons; % activated was higher in the top vs bottom decile (Syt10 p = 0.007, Th p = 0.012). Syt10 led both conditional probabilities and a mixed model controlling for position and ChrimsonR signal; Syt10+ cells out-responded low-expressing neighbours (median 45.3 µm away). · Spinoparabrachial input recruits a molecularly…; Regression against neuron location and ChrimsonR…; Comparison with neighboring neurons; Figure legends

4. 04

   ShownIn awake mouse visual cortex, coupling to pupil-linked arousal differed across Pvalb, Sst and Vip interneurons recorded at the same time. Calb2+ VIP neurons were more positively coupled than Calb2- VIP neurons.

   Pvalb+ cells split roughly evenly between positive and negative pupil correlation, Vip+ cells were mostly positive and Sst+ cells were intermediate (within-mouse permutation Kruskal-Wallis H = 190.79, p < 0.001, n = 2,385 cells; Friedman χ² = 11.10, p = 0.011, 4 mice). Calb2+ minus Calb2- mean r: VIP 0.092 (p < 0.001, 680 neurons), SST 0.062 (p = 0.162, 600 neurons), 4 mice. · Distinct arousal tuning and spatial correlations…; Statistical testing; Figure legends

5. 05

   SuggestedCorrelations between neurons of the same class fell off with distance most slowly for VIP, at an intermediate rate for SST and steeply for PV and putative excitatory neurons. Calb2+ SST and VIP subtypes had longer decay lengths, and longer decay went with stronger arousal coupling mainly in VIP cells.

   The decay constant λ differed across classes (H = 107.10, p < 0.001, n = 1,864 neurons; Friedman χ² = 12.00, p = 0.007, 4 mice), but the fitted amplitude did not (H = 5.16, p = 0.082). Calb2+ vs Calb2- Δmean log λ: SST 0.386, VIP 0.638 (both p < 0.001; 456 and 360 neurons, 3 mice). Within each class, the λ-arousal correlation was weaker in PV and SST and reversed sign in unlabeled neurons. · Distinct arousal tuning and spatial correlations…; Spatial correlation decay; Statistical testing; Figure legends