Statistical Connectomics

Thesis Committee Meeting

Jaewon Chung

(he/him) - NeuroData lab
Johns Hopkins University
Department of Biomedical Engineering

j1c@jhu.edu
@j1c (Github)
@j1c (Twitter)

Outline

• What we've done

  • Deriving Connectomes of Human Brains
  • Statistical Modeling for Connectomes
  • Heritability of Human Connectomes
  • graspologic + hyppo + m2g

• Graduation plan

Representing brains as networks

Connectomes from diffusion MRI (dMRI)

• in vivo imaging technique
• Exploits direction of water diffusion
  • Anisotropic in white matter tracts
  • Isotropic in other tissues
• Estimates of number of white matter tracts via tractography

MRI to graphs (m2g)

Outline

• What we've done

  • Connectomes of Human Brains
  • Statistical Modeling for Connectomes
  • Heritability of Human Connectomes
  • graspologic + hyppo + m2g

• Graduation plan

Different data, same statistics (Ascombe's Quartet)

• These four datasets have same statistics!

Different networks, same statistics

• These four networks have same (graph) statistics!

Statistical models for networks

• Random dot product graphs (RDPGs)
  • Each vertex has a low dimensional latent position.
  • Estimate latent position matrix via adjacency spectral embedding.
  • =

Two sample graph testing

Outline

• What we've done

  • Connectomes of Human Brains
  • Statistical Modeling for Connectomes
  • Heritability of Human Connectomes
  • graspologic + hyppo + m2g

• Graduation plan

Heritability of connectomes?

• Heritability = proportion of phenotypic variance due to genetic variance
  • Predict disease rick
  • Potential for targeted treatments
  • Genes -> structure -> function -> behavior

Heritability as causal problem

Do genomes affect connectomes?

Do genomes affect connectomes given covariates?

Human Connectome Project

• Brain scans from identical (monozygotic), fraternal (dizygotic), non-twin siblings.
• Regions defined using Glasser parcellation (180 regions).

Methods of comparing connectomes

• Exact : measures all differences in latent positions
  • Differences in the latent positions implying differences in the connectomes themselves
• Global : considers the latent positions of one connectome are a scaled version of the other
  • E.g. males may have globally fewer connections than females
• Vertex : similar to the global differences, but it allows for each vertex to be scaled differently
  • E.g regions have preferences in connections
  • regions tend to connect strongly within hemisphere

We see stochastic ordering along familial relationships

We detect heritability (associational)

Some signals disappear after conditioning

To sum up...

• Statistical models = nuanced investigations
• Connectomes are dependent on genome, up to some common structures.

Outline

• What we've done

  • Connectomes of Human Brains
  • Statistical Modeling for Connectomes
  • Heritability of Human Connectomes
  • graspologic + hyppo + m2g

• Graduation plan

How to use these tools?

Outline

• What we've done

  • Connectomes of Human Brains
  • Statistical Modeling for Connectomes
  • Heritability of Human Connectomes
  • graspologic + hyppo

• Graduation plan

Summary of work so far

Summary of work to be done

Graduation May 2024

Acknowledgements

Team

NeuroData lab, Microsoft Research

Feedback?

Jaewon Chung

j1c@jhu.edu
@j1c (Github)
j1c.org

Appendix

How do we compare genomes?

• Neuroimaging twin studies do not sequence genomes.
• Coefficient of kinship ()
  • Probabilities of finding a particular gene at a particular location.
• d(Genome, Genome) = 1 - 2.

Neuroanatomy (mediator), Age (confounder)

• Literature show:
  • neuroanatomy (e.g. brain volume) is highly heritable.
  • age affects genomes and potentially connectomes
• d(Covariates, Covariates) = ||Covariates - Covariates||

How do we compare connectomes?

• Random dot product graph (RDPG)

  • Each vertex (region of interest) has a low dimensional latent vector (position).
  • Estimate latent position matrix via adjacency spectral embedding.

• d(Connectome, Connectome) =

Distance correlation

• Measures dependence between two multivariate quantities.
  • For example: connectomes, genomes.
• Can detect nonlinear associations.
• Measures correlation between pairwise distances.

Conditional distance correlation

• Augment distance correlation procedure with third distance matrix.

Associational Test for Connectomic Heritability

• Connectome, GenomeConnectomeGenome
  Connectome, GenomeConnectomeGenome

Associational Test for Neuroanatomy

• Neuroanatomy, GenomeNeuroanatomyGenome
  Neuroanatomy, GenomeNeuroanatomyGenome

Causal Test for Connectomic Heritability

• Conn., Genome|CovariatesConn.|CovariatesGenome|Covariates
  Conn., Genome|CovariatesConn.|CovariatesGenome|Covariates