Module 20: Statistical Models and Inference for Connectomics

Lesson Flow

Learn

Goals and Concepts

Start with the capability target and concept set for this module.

Practice

Studio Activity

Apply the ideas in a guided activity tied to realistic outputs.

Check

Assessment Rubric

Use the rubric to verify competency and identify improvement targets.

Teach

Slides and Worksheets

Open the teaching deck, worksheet, and editable slide source.

Interactive Lab

Practice in short loops: checkpoint quiz, microtask decision, and competency progress tracking.

Checkpoint Quiz

Q1. Which output most clearly demonstrates module competency? A measurable artifact linked to an explicit method A general reflection about interest in neuroscience A list of future goals without evidence

Competency is shown through measurable, method-linked evidence.

Q2. What should always accompany a technical claim in this curriculum? A limitation or uncertainty statement A motivational quote A citation count

Every claim should include boundaries and uncertainty.

Q3. What is the best next step after identifying a gap in understanding? Define a concrete practice task and verification criterion Skip to a harder module Ignore the gap if progress is slow

Progress improves when gaps become explicit practice targets.

Microtask Decision

Choose the action that best improves scientific reliability.

Record method, version, and decision rationale before sharing results Share results quickly and document details later Rely on memory for why decisions were made

Progress Tracker

State is saved locally in your browser for this module.

Read capability target
Complete studio activity
Review assessment rubric
Pass checkpoint quiz
Complete microtask
Complete annotation challenge

0% complete

Annotation Challenge

Click the hotspot with the strongest evidence for the requested feature.

Selected hotspot: none

Capability target

Design and execute a connectomics inference plan that includes null-model choice, multiplicity control, uncertainty reporting, and explicit claim boundaries.

Why this module matters

Connectomics analyses can produce thousands of statistically testable patterns. Without disciplined inference, teams risk publishing artifacts from preprocessing bias, multiple comparisons, or misaligned null assumptions.

Concept set

1) Null models encode scientific assumptions

Technical: null models should preserve relevant graph constraints (degree sequence, spatial limits, cell-class composition) while randomizing the tested structure.
Plain language: your “chance baseline” must reflect biology and data collection realities.
Misconception guardrail: a generic random graph is rarely an adequate connectomics null.

2) Multiplicity is structural, not optional

Technical: motif families and subgroup analyses require correction strategies and predeclared test hierarchies.
Plain language: if you test many patterns, some will look significant by accident.
Misconception guardrail: reporting only p-values without multiplicity context is incomplete.

3) Exploratory and confirmatory analyses must be separated

Technical: hypothesis generation and hypothesis testing should have different reporting labels and evidence standards.
Plain language: be clear about what you discovered versus what you validated.
Misconception guardrail: post-hoc storytelling is not confirmatory inference.

4) Statistical challenges unique to connectomics

Connectomics datasets present several statistical difficulties that are uncommon in other fields. Massive multiple comparisons arise when testing thousands of motifs, cell-type pairs, or connection patterns simultaneously. Spatial autocorrelation is pervasive because nearby neurons share arbor overlap, creating non-independent edges that violate standard test assumptions. The threshold problem is particularly acute: choosing a minimum synapse count (e.g., 3 vs. 5 synapses to define a “real” connection) changes the resulting graph and all downstream statistics, yet no universally accepted threshold exists.

Researcher degrees of freedom in null model selection further compound these issues. Different null models that preserve different graph properties (degree sequence, spatial distance distribution, cell-type composition) can yield contradictory conclusions from the same data. Best practices include using permutation tests over parametric alternatives when distributional assumptions are uncertain, reporting effect sizes alongside p-values to distinguish statistical significance from biological relevance, and performing sensitivity analyses across multiple thresholds and null model variants to confirm that findings are robust rather than artifacts of a single analytical choice.

Core workflow: connectomics inference protocol

Question-to-test mapping
- Convert biological question into estimand(s), test set, and effect-size target.
Null-model design
- Define null constraints and why they preserve key confounders.
Inference execution
- Run model/tests with preregistered thresholds and multiplicity controls.
Robustness checks
- Test sensitivity to preprocessing variant, sampling region, and parameter choice.
Claim calibration
- Report supported, uncertain, and unsupported claims in separate blocks.

Studio activity: motif inference challenge

Scenario: A team reports motif enrichment in one dataset and asks whether the claim generalizes.

Tasks

Propose at least two candidate null models and justify each.
Run or outline multiplicity-aware testing strategy across motif set.
Draft a results summary separating exploratory and confirmatory findings.
Add one robustness check for cross-dataset comparability.

Expected outputs

Inference design sheet (estimand, null, tests, correction).
One-page claim calibration summary.
Robustness plan with pass/fail criteria.

Assessment rubric

Minimum pass
- Null model is justified and constraints are explicit.
- Multiplicity handling is documented and applied.
- Claims are partitioned by confidence level.
Strong performance
- Demonstrates sensitivity analysis against preprocessing and sampling choices.
- Reports effect sizes and uncertainty, not significance alone.
- Provides clear boundaries on generalization.
Common failure modes
- Null model choice disconnected from biological question.
- Selective reporting of significant outcomes.
- Conflation of exploratory signal with validated inference.

Content library references

Motif analysis — Null models and statistical testing for motifs
Network analysis methods — Graph metrics requiring statistical interpretation
Metrics and QA — Quality metrics with statistical properties

Teaching resources

Core unit context: Connectome Analysis and NeuroAI
Reading support: Technical Track Journal Club
Dataset workflow context: Workflow overview
Quality controls context: Connectome Quality tool

Evidence anchors from connectomics practice

Key papers to use in this module

Key datasets to practice on

Competency checks

Can you defend your null-model assumptions in one paragraph?
Can you report one finding with effect size, uncertainty, and limitation?
Can you identify which result remains exploratory?

Quick practice prompt

Write a 6-8 sentence inference note that includes:

hypothesis and estimand,
null-model assumptions,
multiplicity strategy,
one robust conclusion and one unresolved uncertainty.

Teaching Materials

Slide Deck

Classroom-ready deck links for teaching and delivery.

Open slide deck page

Open rendered HTML deck

Download PowerPoint (.pptx)

Activity Worksheet

Learner worksheet aligned to the studio activity and rubric.

Open worksheet

Slide Source

Marp source file for editing and rendering.

course/decks/marp/modules/module20.marp.md