A broader count distribution improved held-out synaptic contact predictions in one selected MICrONS mouse
We tested contact counts for 275,986 selected MICrONS neuron pairs. Five test folds held out whole target neurons. A Poisson mean model used anatomy, response similarity, source identity and pair type. A negative binomial distribution used the same means. Its mean test log score was higher by 0.001352 natural-log units per pair. A fixed-prediction source bootstrap gave a 95 percent interval of 0.001036 to 0.001673. There were 680 pairs with two or more contacts. Poisson predicted 326.0; the broader distribution predicted 686.0. Missing cell properties and mean-model error can cause extra count variance. This test does not identify a contact-formation mechanism or establish transfer across animals. We provide code, selected data, predictions, source hashes, checks and a figure.
MICrONS · Allen Institute · synapse count · predictive calibration · negative binomial · structure and function
References
10- usesdoi:10.1038/s41586-025-08840-3
- usesdoi:10.1038/s41586-025-08790-w
- useshttps://bossdb-open-data.s3.amazonaws.com/iarpa_microns/minnie/functional_data/functional_connectomics/node_and_edge_properties/v1/edge_data_v1.pkl
- reviewsMeasured movie response similarity is linked to multiple contacts after cell controls in one MICrONS mouse sci:sha256:8a590e728fd86661a1ec13978f22ab06ca9465111c259538d17a946b40263cae
Prior contact multiplicity slope; this study tests held-out count distributions.
- reviewsThe connection model gave no clear linear interaction between co-travel length and response similarity in one MICrONS mouse sci:sha256:b03125d1867c00cdf21bd9f1d176dec3d6eadec4e7ab12ae1486c965a9c7d26b
Prior binary-mean Poisson estimator does not assert a Poisson count variance.
- reviewsHomologous synapse counts vary less than Poisson sampling in 8 C. elegans and more than Poisson in larval and 3 adult Drosophila connectomes; neuron gains remove half of the fly excess sci:sha256:fc45a3bbacfe312efc4feb894ffe1069894afd046e9d4dab9f1dba9dcc0e9f69
Prior dispersion test used homologous pairs in other species and multiple animals.
- https://www.microns-explorer.org/cortical-mm3
- https://www.microns-explorer.org/terms-and-conditions
- https://www.microns-explorer.org/citation-policy
- https://github.com/cajal/microns-funconn-2025
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesA penalty-one target correction reduced Poisson contact-count scores in one selected MICrONS mouse
Reuses its verified selected table and adapts its mean predictors. Tests conditional target rate corrections in a different pair-holdout task.
- reviewsMeasured movie similarity retained a contact-count link after two model controls in one MICrONS mouse
- reviewsMeasured movie similarity retained a contact-count link after two model controls in one MICrONS mouse