Session profile
- Audience: advanced learners bridging connectomics with computational modeling.
- Duration: 85 minutes lecture + 15 minutes motif-design exercise.
- Output: executable motif-analysis plan with null-model justification.
Slide-by-slide lecture plan
- Slide 1 (2 min): Title and learning objective
- Slide 2 (5 min): From reconstructed circuit to testable graph hypotheses
- Slide 3 (6 min): Motif analysis workflow overview
- Slide 4 (6 min): Query language and representation choices
- Slide 5 (6 min): Subgraph isomorphism complexity and tooling
- Slide 6 (6 min): Null models I
- degree preserving and random rewires.
- Slide 7 (6 min): Null models II
- spatial and cell-type constrained controls.
- Slide 8 (7 min): Multiple-testing and statistical interpretation
- Slide 9 (7 min): Worked query example (DotMotif-style)
- Slide 10 (7 min): Reproducibility requirements
- versioned data, query code, seeds.
- Slide 11 (5 min): Cross-dataset comparability caveats
- Slide 12 (5 min): NeuroAI transfer: where it helps, where it overreaches
- Slide 13 (6 min): Failure modes
- post-hoc hypotheses, null mismatch, overgeneralization.
- Slide 14 (6 min): Activity + debrief
- define motif, null, and success criterion.
- Primary shortlist:
course/units/figures/09-connectome-analysis-neuroai-selected-v1.md.
- Use one algorithmic complexity visual, one motif query visual, one benchmark caveat panel.
Speaker notes (expert-level)
- Keep exploratory and confirmatory analyses explicitly separated.
- Treat all benchmark performance from older decks as historical unless revalidated.
Assessment and artifacts
- Deliverable: motif-analysis protocol card.
- Rubric dimensions: hypothesis clarity, statistical rigor, and reproducibility.
Connections
Slide source file
- Marp draft source:
course/decks/marp/09-connectome-analysis-neuroai.marp.md
- Batch render helper:
./scripts/render_marp.sh
