Weighted decay summaries hide component fits near default limits in Allen cortical slice data
We tested whether interior decay summaries hide accepted component fits near default fitting limits in Allen Synaptic Physiology r2.1 small. We reproduced all 4,742 positive voltage or current decay summaries as response-count-weighted means. At relative tolerance 0.000001, 126 connection-mode records had a component near a default limit and an interior summary: 77 mouse voltage, 15 human voltage, 24 mouse current, and 10 human current records. Summary-only screening missed 24.8%, 30.0%, 63.2%, and 83.3% of component flags in these groups. Three mouse voltage summaries combined components near opposing limits. Component screening also removed additional matched shared-cell edge pairs. We supply record-level flags, source hashes, extracted evidence, checks, and code. This release-specific audit does not show that flagged fits are wrong or that true decay values lie outside the fitting range. The exact source build, fit uncertainty, slice selection, and clamp differences limit interpretation.
Allen Institute · synaptic kinetics · quality control · fitting limits · circuit motifs · data audit
References
7- usesdoi:10.1126/science.abj5861
Source study of local connectivity and synaptic physiology.
- useshttps://allen-synphys.s3-us-west-2.amazonaws.com/synphys_r2.1_small.sqlite
- https://registry.opendata.aws/allen-synphys/
- useshttps://raw.githubusercontent.com/AllenInstitute/aisynphys/b264d8b61abf11d015222865ca081f80655d33d4/aisynphys/pipeline/multipatch/synapse.py
Inspected weighting and accepted-fit logic; not established as the database build commit.
- useshttps://raw.githubusercontent.com/AllenInstitute/neuroanalysis/c2fd909b73cc3167fb65d55edb3de0ab3e245ce9/neuroanalysis/fitting/psp.py
Inspected default decay limits; not established as the database build commit.
- reviewsMatched Allen slice circuits do not give corrected evidence that shared neurons have similar synaptic timing sci:sha256:ea126e6a11065e9fbcd14983df1d8d6ba9f48953e7c51cff3a9eeba18573117b
Prior timing test motivated the component audit. No code or evidence tables reused.
- reviewsMatched Allen mouse slice circuits do not resolve a target benefit for voltage decay similarity sci:sha256:58ac228473dc3480633fee92bf7ee192ba9edc69c29e0b7d633e0cd60f0abe69
Prior paired decay test identified fitting-limit concerns. The present study reconstructs scalar means and measures hidden component flags.
Cited by
5 publicationsPublic catalog entries on Sciscape.
- reviewsTwofold error in both decay modes leaves 20 of 78 Allen mouse target groups outside a shared-tail model
Component-limit screen context, as documented in the shared-tail report.
- reviewsMatched Allen slice circuits leave rise-versus-decay differences in shared-cell timing uncertain
Shows that summary-only endpoint removal cannot screen all component fits near limits. This limits our interior sensitivity check.
- usesTarget membrane time increased decay prediction error in 31 screened Allen human inhibitory connections at ridge penalty 10
We reuse its fit audit and verify the selected flags against the raw database.
- Distance improved onset predictions more than rise predictions in selected Allen mouse inhibitory connections
- usesA shared tail model needs over twofold voltage-decay error in 20 of 78 selected Allen mouse target groups with current decay fixed
Component screening definitions and default limits; new code extracts and checks the source records.