Removing one of four mouse records reduced the mean Gabazine-coded change in retinal response correlation from 0.100 to -0.002

We tested if paired retinal drug data give a clear test of opposite roles for GABA and glycine inhibition. We used the Franke et al. processed files and paired control and drug traces by their common ROI row. Of 1,799 drug-coded release regions, 903 had all four finite chirp traces. Requiring both control quality indices to be at least 0.3 kept 802 regions from seven distinct date-and-mouse keys. We calculated local versus wide-field Pearson correlation from 2 to 32 seconds. We averaged regions within scans, scans within mice, and then mice. For code 1, named Gabazine in the source guide, the mean correlation change was 0.100 across four mouse keys. Excluding one key gave -0.002. The exact two-sided sign-flip reference was 0.625. Codes 2, 3, and 4 gave mean changes of -0.676, -0.537, and -0.814, but had only two, one, and one mouse keys. The guide lists Gabazine and TPMPA separately, while the paper describes combined GABA blockade. We retain the stored labels and cannot resolve that mapping. These data do not give a clear mouse-level test of the proposed circuit distinction. Separate trace normalization also prevents an absolute gain test. The contribution is a checked native-grid subset, coverage tables, mouse-level and time-window checks, code, figures, source hashes, and repeat instructions. It does not claim discovery of the inhibitory mechanism.

retina · bipolar cells · structure and function · pharmacology · GABA · glycine · data provenance · mouse-level analysis