+ - 0:00:00
Notes for current slide
Notes for next slide

Analyzing high dimensional metabolic data in epidemiological studies

Luke Johnston

Aarhus University

1 / 27
Aarhus University

Aims:

  • Describe NetCoupler

  • Get feedback and/or comments on it

2 / 27
Aarhus University

Aims:

  • Describe NetCoupler

  • Get feedback and/or comments on it

  • ... and to see if I explained it well

2 / 27
Aarhus University

THE PROBLEM

3 / 27
Aarhus University

Modern epidemiology studies generate a lot of metabolic data

  • -omics type data

  • Metabolic biomarkers

  • High dimensionality

  • Complex networks

4 / 27

Like metabolomics, fatty acid panels.

Classic epidemiology meaning having an exposure with an output

Aarhus University

"Traditional" analysis may use: Dimensionality reduction

Reducing number of variables with PCA

5 / 27

This has the advantage of making things simpler while trying to maximize variance in the data. Afterward you can do modelling on each principal component. The disadvantage of this approach is that it loses a lot of information since the interdependence and connections between variables it not maintained.

Aarhus University

"Traditional" analysis may use: Many regression-type models

O1 = M1 + covariates
O1 = M2 + covariates
...
O1 = M7 + covariates
O1 = M8 + covariates
6 / 27

Some ways you might go about analyzing this data is by running many regression models, one for each metabolic variable for instance.

This of course has problems since you're simply running a bunch of models and not taking account of the inherent interdependencies between variables.

Aarhus University

"Traditional" analysis may use: Network analysis

7 / 27

This approach is nice in that you can extract information about the connection between metabolic variables. But there is no way to incorporate the disease outcome with this approach and in order to construct the network properly most methods require you provide a prespecified base network, which you might not know.

Aarhus University

So, what if we...

  • want info about network structure?
8 / 27
Aarhus University

So, what if we...

  • want info about network structure?
  • don't know the network structure?
8 / 27
Aarhus University

So, what if we...

  • want info about network structure?
  • don't know the network structure?
  • have an exposure, metabolites, and outcome?
8 / 27
Aarhus University

So, what if we...

  • want info about network structure?
  • don't know the network structure?
  • have an exposure, metabolites, and outcome?
  • are interested in causal links?
8 / 27
Aarhus University

What if we want to know something like this?

9 / 27
Aarhus University

THE SOLUTION

10 / 27
Aarhus University

NetCoupler: A "causal structure learning" algorithm that...

  • Finds most likely network structure

  • Allows inclusion of exposure and outcome

  • Identifies causal links between and within network

NetCoupler was developed by Clemens Wittenbecher.

11 / 27
Aarhus University

Algorithm (steps) used in NetCoupler

12 / 27
Aarhus University

1. DAG skeleton of network is estimated

13 / 27
Aarhus University

14 / 27
Aarhus University

3. Direct effects estimated between exposure and network

15 / 27
Aarhus University

16 / 27

If you were interested in linkages with an exposure on the network, you could stop here.

Aarhus University

17 / 27
Aarhus University

6. Direct effects estimated between network and outcome

18 / 27
Aarhus University

19 / 27

You could stop here if you were interested in linkages with the network and an outcome.

Aarhus University

8. (optional) Combine both estimations into joint model

20 / 27
Aarhus University

Infer (potentially) causal pathways with graphical model

21 / 27
Aarhus University

OUR WORK

22 / 27
Aarhus University

Current state of NetCoupler

  • Series of R scripts
23 / 27
Aarhus University

Current state of NetCoupler

  • Series of R scripts

  • Only uses specific methods

23 / 27
  • Greats well on Clemens' computer

  • Not readily re-usable by others

  • PC-algorithm for network

  • Cox for modelling
  • Not easily extensible to other models
Aarhus University

Converting to a usable R package

GitHub repo

24 / 27
Aarhus University

Converting to a usable R package

Repo project board

25 / 27

We've got to improve on the underlying implementation, optimize the code for better speed and performance, fix up the interface so that other methods can be used like regression or mixed models, and reduce some software dependencies since right now it relies on software that is really hard to install and get working.

Aarhus University

Goals:

  • Submit to CRAN[1]
26 / 27
Aarhus University

Goals:

  • Submit to CRAN[1]

  • Create simple interface to use

26 / 27
Aarhus University

Goals:

  • Submit to CRAN[1]

  • Create simple interface to use

  • Create website with documentation

26 / 27
Aarhus University

Goals:

  • Submit to CRAN[1]

  • Create simple interface to use

  • Create website with documentation

  • Develop tutorials

[1] CRAN distributes R packages.

26 / 27
Aarhus University

Interest? Feedback? Comments?

27 / 27
Aarhus University

Aims:

  • Describe NetCoupler

  • Get feedback and/or comments on it

2 / 27
Paused

Help

Keyboard shortcuts

↑, ←, Pg Up, k Go to previous slide
↓, →, Pg Dn, Space, j Go to next slide
Home Go to first slide
End Go to last slide
Number + Return Go to specific slide
b / m / f Toggle blackout / mirrored / fullscreen mode
c Clone slideshow
p Toggle presenter mode
t Restart the presentation timer
?, h Toggle this help
Esc Back to slideshow