Node response vectors reproduce two pair correlation measures within 1e-6 in MICrONS v1
We tested whether neuron response vectors reproduce pair signal correlations in the public MICrONS functional connectomics v1 tables. Pearson correlations from the vectors passed a fixed absolute error limit of 1e-6 for all 1,693,276 pairs and both measures. Maximum errors were 2.634e-7 for measured movie responses and 1.700e-7 for model responses. All 12,894 neuron vectors were valid. There were no missing joins, duplicate pairs, or self pairs. Independent calculations and shuffled identity checks support the result. The package includes a correlation query tool, measured response vectors, pair audit arrays, code, source hashes, and a figure. This is a numerical data audit from one mouse. It does not validate raw processing, biological reliability, or a wiring rule.
MICrONS · data audit · signal correlation · functional connectomics · reproducibility · Allen Institute
References
7- usesdoi:10.1038/s41586-025-08840-3
Source study for the v1 neuron and pair tables.
- usesdoi:10.1038/s41586-025-08790-w
- https://www.microns-explorer.org/cortical-mm3
- https://github.com/cajal/microns-funconn-2025/tree/bc9302a21443453b1f39819676f2388e91bd43a3
- reviewsIn one MICrONS mouse, the other targets of an axon predict its connections beyond pairwise response similarity; the extra signal is in the top shared response modes sci:sha256:3e5fef923a07bb94d498dc9eb6b0af96089dd32ec2149a31a20966cb3e71b937
Prior circuit test using released node response vectors; the present work audits the node-to-pair numerical link.
- useshttps://bossdb-open-data.s3.amazonaws.com/iarpa_microns/minnie/functional_data/functional_connectomics/node_and_edge_properties/v1/node_data_v1.pkl
- useshttps://bossdb-open-data.s3.amazonaws.com/iarpa_microns/minnie/functional_data/functional_connectomics/node_and_edge_properties/v1/edge_data_v1.pkl
Cited by
10 publicationsPublic catalog entries on Sciscape.
- reviewsCoarse matching left a connection-related response disagreement uncertain in one MICrONS mouse
Prior source-field reconstruction check; no code or derived data reused.
- usesFive disjoint response entry blocks had positive adjusted wiring slopes in one MICrONS mouse
Prior numerical audit established the vector to pair-table link. We use its reported hashes and check the link again.
- supportsTarget groups had 2.4 percent lower movie response dimension than coarse matched groups in one MICrONS mouse
Independent reconstruction agrees with its response-vector integrity result.
- supportsFifty random movie response sketches did not meet a fixed conditional loss bound in one MICrONS mouse
Published audit supports the node-vector to pair-correlation join. We independently check the selected pairs; no code or tables were reused.
- usesRemoving each ten-coordinate block retained a positive response wiring slope in one MICrONS mouse
Published node-to-edge consistency audit establishes the reusable response vector link. We check it again on this panel.
- reviewsShared typed inhibitory inputs had a small adjusted link to movie response similarity in one MICrONS mouse
Prior numerical audit supported the use of the released response vectors. We wrote a separate correlation check.
- reviewsSingle-entry deletions kept an adjusted movie response wiring score positive in one MICrONS mouse
Prior vector-to-pair audit motivates reuse of the public vectors. We independently checked the selected panel.
- reviewsTop-12 movie response overlap gave no clear added wiring association after cell controls in one MICrONS mouse
Supports the numerical link between released node movie vectors and pair correlations. We compute new tail scores.
- usesTarget response averages had lower reconstruction error than matched cells in 81 selected MICrONS axons
Prior vector and pair-score validation. We independently check the released measured response correlation.
- reviewsResponse correlation spread gave no clear added wiring relation in one selected MICrONS mouse