| --- |
| language: |
| - en |
| - zh |
| task_categories: |
| - other |
| tags: |
| - neuroscience |
| - drosophila |
| - connectomics |
| - kenyon-cells |
| - computational-neuroscience |
| - reproducibility |
| pretty_name: Synapse-resolved KC models and public-data validation |
| --- |
| # Synapse-resolved KC models and public-data validation |
|
|
| Versioned addition to the existing `cyd0806/neuronmimic-data` dataset. The original dataset root and historical releases are not replaced. |
|
|
| This payload covers all nine completed task rounds, the full 5,177-cell structural atlas, transitive data dependencies, predictions, condition traces, quality-control tables, source-data reconstructions and figure/report exports. `release_inventory.csv` is the full local accounting; its `public_disposition` field explicitly distinguishes included files, credential exclusions (if any) and original-source retrieval requirements. `archive_manifest.json` gives archive SHA256, bytes and member counts. Archive members preserve original workspace-relative paths. |
|
|
| The manuscript and Chinese method/conclusion note are supplied under `manuscript_release/`. The intended code repository is https://github.com/ydchen0806/bio_fly_cyber_lateralization on `research/wu-structure-function-20261104`. `publication_status.json` is authoritative for whether a remote publication actually occurred. |
|
|
| ## Scientific scope |
|
|
| The analysis preserves a frozen 32/15/14 exploratory KC split, real trees, pre/post direction, synapse locations and partner identities. Structural joint-feature gains were largely reproduced by matched conditional nulls. The model did not establish natural multisensory synergy or subtype-resolved physiological validation. The public physiology benchmark is a retrospective within-study Fly-ID holdout; the source recordings do not provide per-cell gammaKC/KCab-p identity. The 82 source rows are cells, not 82 verified independent animals. |
|
|
| ## Content and attribution |
|
|
| Derived outputs and attributed source materials retain their original provenance. See `THIRD_PARTY_TERMS.md`, `source_retrieval_ledger.csv` and the source repository/index files within each archive. No uniform MIT or CC BY license is asserted over all third-party code and data. Formal JP288790 Dataset S1 is CC BY 4.0 as documented in the original article XML; original author code without an explicit license is inventoried for original-source retrieval instead of relicensed. |
|
|
| No newly collected animal data are included. Virtual environments, local account credentials, git internals and internal group-meeting records are excluded from scientific public payloads. Failed/superseded scientific runs are retained where required to audit implementation choices and are identified by their original status. |
|
|
| ## Use |
|
|
| Download the manifests first. Verify an archive's SHA256 before extracting it into a fresh workspace. Obtain any original-source retrieval entries separately with their recorded commit and expected SHA256. Install the recorded Python/NEURON/Arbor dependencies for the relevant round. A public file's existence does not mean its contents are a new biological validation dataset. |
|
|
| Contact: ydchen0806; yindachen@mail.ustc.edu.cn. |
|
|
