1 Introduction

1.1 Overview

This tutorial will focus on the application of two multivariate Bayesian software programs: (1) bnlearn and (2) mvBIMBAM.

Approach 1: bnlearn - bnlearn is an R package for learning the graphical structure of Bayesian networks, estimating their parameters, and performing some useful inference as described in/at:

Scutari M, Howell P, Balding DJ, Mackay I. Multiple Quantitative Trait Analysis Using Bayesian Networks. Genetics. 2014 Apr 11;198(1):129–137. PMID: 25236454 PMCID: PMC4174925 DOI: https://doi.org/10.1534/genetics.114.165704

Zhang JZ, Heinsberg LW, Krishnan M, Hawley NL, Major TJ, Carlson JC, Harré Hindmarsh J, Watson H, Qasim M, Stamp LK, Dalbeth N, Murphy R, Sun G, Cheng H, Naseri T, Reupena MS, Kershaw EE, Deka R, McGarvey ST, Minster RL, Merriman TR, Weeks DE. Multivariate analysis of a missense variant in CREBRF reveals associations with measures of adiposity in people of Polynesian ancestries. Genet Epidemiol. 2023 Feb;47(1):105-118. doi: 10.1002/gepi.22508. Epub 2022 Nov 9. PMID: 36352773; PMCID: PMC9892232.

www.bnlearn.com

Additional details about the method and its interpretation can be found in the bnlearn chapter of this book or 01_mvNUR_bnlearn.Rmd in the GitHub repository.

Approach 2: mvBIMBAM - mvBIMBAM implements a terminal-based Bayesian approach for genetic association analysis of multiple related phenotypes, as described in:

Shim H, Chasman DI, Smith JD, Mora S, Ridker PM, Nickerson DA, Krauss RM, Stephens M. A multivariate genome-wide association analysis of 10 LDL subfractions, and their response to statin treatment, in 1868 Caucasians. PLoS One. 2015 Apr 21;10(4):e0120758. doi: https://doi.org/10.1371/journal.pone.0120758. PMID: 25898129; PMCID: PMC4405269.

Stephens M. A unified framework for association analysis with multiple related phenotypes. PLoS One. 2013 Jul 5;8(7):e65245. doi: https://doi.org/10.1371/journal.pone.0065245. Erratum in: PLoS One. 2019 Mar 19;14(3):e0213951. PMID: 23861737; PMCID: PMC3702528.

Additional details about the method and its interpretation can be found in the mvBIMBAM chapter of this book or in 02_mvNUR_mvBIMBAM.Rmd in the GitHub repository.

Note that while bnlearn is run directly in R, mvBIMBAM is a terminal-based program with no point/click desktop app. As such, please note that this example code, particularly the mvBIMBAM approach, requires at least an introductory understanding of R and Unix (terminal-based computer navigation).

Finally, note that, throughout all markdowns, we have flagged lines that will need to be modified by the user if adapting this code for your own data using ###CUSTOMIZE** annotation.

1.2 Installation

1.2.1 bnlearn

bnlearn is available on CRAN and can be downloaded from its web page in the Packages section (here).

For example, you can install bnlearn from CRAN using:

install.packages("bnlearn")

install.packages("https://www.bnlearn.com/releases/bnlearn_latest.tar.gz", repos = NULL, type = "source")

Development snapshots, which include bugfixes that will be incorporated in the CRAN release as well as new features, can be downloaded from the links above or installed with a simple:

install.packages("http://www.bnlearn.com/releases/bnlearn_latest.tar.gz")

The only suggested packages not hosted on CRAN are graph and Rgraphviz, which can be installed from BioConductor:

if (!requireNamespace("BiocManager", quietly = TRUE))
  install.packages("BiocManager")
BiocManager::install()
BiocManager::install(c("graph", "Rgraphviz"))

following the instructions present on this and this webpage. Please also note that the gRain package, while on CRAN, depends on packages that are on Bioconductor both directly and through the gRbase package, which depends on RBGL:

BiocManager::install()
BiocManager::install(c("graph", "Rgraphviz", "RBGL"))
install.packages("gRain")

Check your R version if you are having any issues installing. As of August 2024, bnlearn requires R version 4.4.0 or higher.

Additional installation and troubleshooting information can be found at the bnlearn website.

1.2.2 mvBIMBAM

mvBIMBAM is designed to be installed from the mvBIMBAM GitHub respository. The binary executable file for Linux is in the https://github.com/heejungshim/mvBIMBAM/bin/ directory while a zip file of the binary executable file, source code, and example input files for Mac is in https://github.com/heejungshim/mvBIMBAM/forMAC directory. NOTE: The software executable is called ‘bimbam’, but we refer to the program as ‘mvBIMBAM’ throughout.

HOWEVER, when developing/testing this tutorial, several of us ran into installation issues for Mac. Ultimately, we had to modify the program’s makefile to enable successful installation. We have alerted the mvBIMABM software maintainer of this issue, but have not yet seen it resolved. As such, we provide a modified Mac source code download to make this process easier for users of this tutorial (allowable under GNU General Public License (GPLv3+)).

To install mvBIMBAM, please follow the instructions below, which were written with the assumption that users have an introductory understanding of navigating a computer using the terminal/Unix commands. If you are new to using Unix commands in the terminal, please visit this how to guide for a brief introduction.

1.2.2.1 Mac

Download BIMBAM_multi_pheno_MAC_M1_modified.zip or BIMBAM_multi_pheno_MAC_Intel_modified.zip from our GitHub repository. The former is for newer Apple silicon/M1 macs and the latter is for older macs. If you are not sure which Mac you have, you can click on the apple symbol in the upper left corner of your machine and choose the “About This Mac”. If you have an Apple silicon machine, the chip will be listed as Apple M1 or M2. If you have the Apple Intel, you will see the processor listed as Intel Core i5, i7, or similar.
Unzip the file by double clicking the zipped folder.
Using the terminal, navigate to the newly unzipped mvBIMBAM folder. For example:

cd Desktop/BIMBAM_multi_pheno_MAC_M1_modified
pwd

cd Desktop/BIMBAM_multi_pheno_MAC_Intel_modified
pwd

For example, on my machine, when I use the pwd command (print working directory), my download folder is located at: /Users/username/Desktop/BIMBAM_multi_pheno_MAC

Run make clean and make all terminal commands which will activate the software build automation tool to compile the C++ source code into an executable bimbam program.

make clean
make all

Note that there is, unfortunately, no obvious sign that install was successful. Installation can be confirmed in step 5.

Move the executable file to system’s PATH environment variable, such as /usr/local/bin, so the program can be called from any location (vs. only the folder where the executable file resides). To do this, navigate to the location of the Unix executable file via the terminal and then use the move (mv) command.

sudo mv bimbam /usr/local/bin/

Confirm that the install/move worked via this command:

which bimbam

For example, on my machine, bimbam is located at /usr/local/bin/bimbam now.

Now if you type:

bimbam

at the Unix prompt and hit return, it should print out this message:

 BIMBAM version 0.99a, visit http://stephenslab.uchicago.edu for possible update.
 Developed by Yongtao Guan ytguan.at.gmail.com, all rights to be lefted to GNU GPL.
 References: Guan and Stephens (2008), Servin and Stephens (2007), Scheet and Stephens (2006).
 Updated March 14 2009.

1.2.2.1.1 Modification details

If you are interested in more details related to the installation issue, and how we modified the makefile to correct this issue, please see the installation hack appendix of this book or Installation Appendix of the GitHub repository.

1.2.2.1.2 Troubleshooting

Note that you need to have command line tools and GSL (GNU Scientific Library) installed on your machine before you can install mvBIMBAM. Please see our troubleshooting appendix of this book or Installation Appendix of the GitHub repository for tips on getting GSL installed on your machine.

1.3 Synthetic data set overview

To facilitate hands-on learning, a synthetic data set was created for use with this example code. The synthetic data set was created from a study of women with breast cancer (BrCa). This data set contains no real data, but mimics the correlation structure of the true data set. The original data set contained data for 110 participants and included data for 6 symptoms, 3 demographic variables, and 6 SNPs. We created a synthetic version of this data set with fake data for 770 participants for use in this tutorial.

Data set name: BrCa_synthetic.csv (see GitHub repository for details).

Variable descriptions:

Symptoms/Outcomes
-ID: Participant ID
-EMO_tscore: Anxiety (PROMIS Anxiety T Score)
-bdito: Depression (Beck’s Depression Inventory)
-FAT_tscore: Fatigue (PROMIS Fatigue Score)
-paohcif: Cognitive Function (Patient’s Assessment of Own Functioning Inventory - Cognitive Function)
-EPSscore: Sleepiness (Epworth Daytime Sleepiness)
-worst_pain: Pain (Self-reported numeric rating of “average pain”)

Demographics/Covariates
-age: Age in years
-education: Education in years
-race: Race (Self-identified, 0=White, 1=Black)

SNPs/Predictors
-rs #s (SNPs)

# Read in a mapping of variable name to informative label
dict <- read.csv("data/BrCa_synthetic_SimpleDict.csv")
pander::pander(dict, "Data dictionary")

Data dictionary (continued below)
Variable	Label	Description
Emo_tscore	Anxiety	PROMIS Emotional Distress Anxiety (Short Form 8a) - 8 items measuring emotional distress and anxiety (panic, fearfulness, worry, dread, tension,nervousness, restlessness, and somatic symptoms including racing heart and dizziness) in the last 7 days. This instrument generates T-scores, which are standard scores with a mean of 50 and standard deviation of 10 in the U.S. general population. Higher T-scores indicate worse distress and anxiety. T-scores range from 10 to 90.
bdito	Depression	The Beck Depression Inventory-II - 21-items measuring severity of depression (i.e., grief, loss, flat affect, withdrawal, mania) over a two-week period. Scores range from 0 to 63, where high scores indicate higher levels or more severe depression.
FAT_tscore	Fatigue	PROMIS Fatigue (Short Form 8a) - 8 items measuring the experience of fatigue and the interference of fatigue on daily activities over the past 7 days. This instrument generates T-scores, which are standard scores with a mean of 50 and standard deviation of 10 in the U.S. general population. Higher T-scores indicate worse fatigue. T-scores range from 10 to 90.
paohcif	Cognitive function	Patient assessment of own cognitive functioning - 33 items measuring self-reported ‘recent’ cognitive function consisting of 5 dimensions: memory, language and communication, use of hands, sensory-perceptual, higher level cognitive and intellectual functioning. Higher scores indicate worse perceived cognitive functioning. Scores range from 0 to 155.
EPSscore	Sleepiness	Epworth Sleepiness Scale - 8 item scale measuring self-reported ‘recent’ daytime sleepiness. Higher score indicate greater daytime sleepiness. Scores range from 0 to 24.
pain	Pain	Brief Pain Inventory (Short Form) - 9 items a widely used clinical tool for assessing pain (worst pain in the last 24 hours [0-10], least pain in the last 24 hours [0-10], pain on average [0-10], paing at the time of the interview [0-10], pain relief from treatment(s) in the last 24 hours [0%-100%]). Higher scores indicate higher levels of pain, each item ranges from 0-10.
rs4880	Variant 1	SOD2 gene, hg19 postion chr6: 160113872
rs5746136	Variant 2	SOD2 gene, hg19 postiion chr6: 160103084
rs1041740	Variant 3	SOD1 gene, hg19 postiion chr21: 33040162
rs10432782	Variant 4	SOD1 gene, hg19 postiion chr21: 33036391
rs4135225	Variant 5	TXN gene, hg19 postiion chr9: 113006691
rs7522705	Variant 6	PRDX1 gene, hg19 postiion chr1: 45992300
race	Race	Self-identified race, 0=White, 1=Black
age	Age	Age in years
education	Education	Self-reported years of education

Type	Citation
Numeric	Pilkonis, P.A. et al. (2011) Item banks for measureing emotional distress from the patient-reported outcomes measurement information system (PROMIS): Depression, Anxiety, and Anger, Assesssment, 18(3), 263-283
Numeric	Beck, A.T. et al. (1996) Beck Depression Inventory-II. San Antonio: The Psychological Corporation
Numeric	Cell, D. et al. (2016) PROMIS fatigue item bank had clinical validity across diverse chronic conditions. Journal of Clinical Epidemiology, 73, 128-134
Numeric	Chelune, G. J. et al. (1986) Neuropsychological and personality correlates of patients’ complaints of disability. Advances in clinical Neuropsychology, 1986:95-126
Numeric	Johns MW. A new method for measuring daytime sleepiness: The Epworth Sleepiness Scale. Sleep 1991; 14(6):540-5
Numeric	Cleeland, C.S. et al. (2009). Pain assessment: Global use of the Brief Pain Inventory. Annals, Academy of Medicine, Sigapore, 23(2), 129-138.
Numeric
Numeric
Numeric
Numeric
Numeric
Numeric
Numeric, encoded value
Numeric
Numeric

1.4 Examples

To prepare the data before running the examples, see the data prep chapter followed by the bnlearn and mvBIMBAM chapters. Note that .Rmd versions of all files are available in our GitHub repository for download or cloning.