NetCoupler: Inferring causal pathways between high-dimensional metabolic data and external factors

Luke W. Johnston

Steno Diabetes Center Aarhus, Denmark

1 / 21

Outline:

History and background on the analytic problem
NetCoupler implementation and development
Examples using NetCoupler
Current challenges

2 / 21

History and background on analytic problem

3 / 21

Amount of data generated is massive

-omics type data
Metabolic biomarkers

High dimensionality
Complex networks

4 / 21

Want to ask questions using this type of data, some way to handle that complexity.

Potential analysis: Dimensionality reduction

Reducing number of variables with PCA.

5 / 21

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.

Potential analysis: Many regression-type models

O1 = M1 + covariates
O1 = M2 + covariates
...
O1 = M7 + covariates
O1 = M8 + covariates

6 / 21

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.

Potential analysis: Network analysis

7 / 21

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.

But, what if...

want info about network structure?

8 / 21

But, what if...

want info about network structure?
don't know the network structure?

8 / 21

But, what if...

want info about network structure?
don't know the network structure?
have an exposure, metabolites, and outcome?

8 / 21

... want to know is something like this?

9 / 21

(Potential) solution: NetCoupler

10 / 21

Initial development

Developed by Clemens Wittenbecher for his PhD thesis
Algorithm that:
- Finds most likely network structure
- Allows inclusion of exposure and outcome
- Identifies causal links from, to, and within the network

Clemens Wittenbecher

11 / 21

Four basic phases of the algorithm

12 / 21

A key is that the algorithm is flexible enough to handle different types of models

Infer (potentially) causal pathways with graphical model output

13 / 21

Developing NetCoupler as an R package

14 / 21

Developing NetCoupler as an R package

github.com/NetCoupler
Goal: Submit to CRAN by mid-2021.

14 / 21

Met him at EDEG, asked if it was an R package. Started working together after that.

Example projects using NetCoupler

15 / 21

UK Biobank: Metabolic pathways between components of stature and HbA1c

16 / 21

UK Biobank characteristics

Basics:
- ~480,000 participants
- Cross-sectional
- Stature measures
- Various demographics

Metabolic variables:
- Cholesterol
- Albumin
- Alanine Aminotransferase
- Apolipoprotein A and B
- Aspartate Aminotransferase
- C-reactive Protein
- Gamma Glutamyltransferase
- HDL and LDL Cholesterol
- Triglycerides

17 / 21

Link between leg length, liver markers, HbA1c

18 / 21

Current challenges

19 / 21

Several limitations or improvements

Mainly, performance can be slow
- E.g. larger data or networks

20 / 21

Several limitations or improvements

Mainly, performance can be slow
- E.g. larger data or networks

Untested on larger networks

20 / 21

Several limitations or improvements

Mainly, performance can be slow
- E.g. larger data or networks

Untested on larger networks
Untested on non-cross-sectional/time-to-event data

20 / 21

Several limitations or improvements

Mainly, performance can be slow
- E.g. larger data or networks

Untested on larger networks
Untested on non-cross-sectional/time-to-event data
Visualizing can be tricky

20 / 21

Several limitations or improvements

Mainly, performance can be slow
- E.g. larger data or networks

Untested on larger networks
Untested on non-cross-sectional/time-to-event data
Visualizing can be tricky
Interpreting estimates can be tricky

20 / 21

Thanks! Questions?

21 / 21

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

Clone slideshow

Toggle presenter mode

Restart the presentation timer

?, h

Toggle this help