Introduction

Purpose

This website allows you to predict disease risk for the individuals of a family

To do this you need to supply

  1. information on the family members and how they are related
  2. A model for the disease for which you want to obtain risks

Both the family (pedigree) information and the disease model can be specified in files and uploaded to this app.

Once the Pedigree and Disease Model are specified, the disease risk for family members can be calculated.

As well as being able to upload your own pedigrees and disease models, you can also select from some example pedigrees and disease models.

Risks can be calculated for a wide range of multifactorial diseases, Mendelian diseases are not covered.

Getting Started

  1. Go to the Pedigree tab. Select the first example from the Example Pedigrees drop down (on left hand side). A diagram of the pedigree should be displayed
  2. Go to the Disease Model tab. Select the first example from the Example Disease Models drop down (on left hand side). It should report the model is valid
  3. Go to the Risk Prediction tab. Click the Calc Risk button (on left hand side)

If successful, a figure and a table will appear on the right hand side of the Risk Prediction page.

The figure shows the predicted risk per pedigree member.

The table is an abbreviated version of the pedigree information, each row representing a family member.

To this table has been added the predicted risk (risk column) for each family member, and also the 5 year risk (nYearRisk column).

This process of specifying the pedigree information and the disese model, then calculating disease risks is the general procedure.

Help on how to specify pedigree information is available in the Help on Pedigrees section of the Pedigrees tab.

Similarly, help on disease model specification is available in the Help on Disease Models section of the Disease Models tab.

Pedigree Info (editable)

This was going to be an editable table but none of the table widgets worked very well


              
              

              

            

          

        

        

          

            

              
            

            

              
Disease Model (editable)

              
Disease Model Parameter values are delineated by whitespace and new lines, and are identified by specific names/headings.

              

              
              

              

            

          

        

        

          

            

              
            

            

              

                

              

            

          

        

        

          
Details

          
This tab reports output generated during risk calculation. It is really just for the developer's user and can be ignored.

          
Upon successful risk estimation the following will be displayed here:

          
    
          
  • the pedigree's various joint distributions
          
  • the disease model parameters assigned to each pedigree member
          
  • pedigree member's risks
          

          
Working directory =  /home/dcampbell/ShinyApps/diseaseRiskPredictorShinyApp

          

            .libPaths() = 
            /home/shiny/R/x86_64-pc-linux-gnu-library/4.4, /usr/local/lib/R/site-library, /usr/lib/R/site-library, /usr/lib/R/library
          

          Error
          

          Warning
          

          Pedigree's Joint tblCalcRiskEst_mPopCov
          

          Pedigree's Joint tblCalcRiskEst_mPopMean
          

          Pedigree's Joint tblCalcRiskEst_mPriorCov
          

          Pedigree's Joint tblCalcRiskEst_mPriorMean
          

          Pedigree's Joint tblCalcRiskEst_mPostCov
          

          Pedigree's Joint tblCalcRiskEst_mPostMean
          

          Pedigree's Joint risk
          

        

        

          
Documentation

          
Methodology

          
This work implements, via a web inteface, the methodology described in Campbell et al. 2010
          (also described in cartoon form here).
          
This work extends that methodology in a number of ways:

          
    
          
  • shared environmental effect can now be modelled. Previous only diseases following an additive genetics and non-shared environment model were modelled.
          
  • the effect of quantitative risk factors can now be modelled. Examples of quantitative risk factors include BMI, blood pressure and polygenic score.
          
  • simplifies modelling of risk factors. Previous the heritability residual to any risk factors had to be calculated.
          
  • Pedigrees that include MZ twins can now be properly modelled.
          

          
Implementation

          
The website was written using the R 'shiny' package (RStudio 2014).

          
Extensive use is made of the 'kinship2' R package (Sinnwell et al. 2014) for pedigree validation and the generation of pedigree diagrams.

          
To improve the risk calculation response time, the Gibbs Sampler used is written in C++ and integrated into the R program using the 'Rcpp' library (Eddelbuettel & Romain 2011).

          
References

          
    
          
  • Campbell, Desmond D., Pak C. Sham, Jo Knight, Harvey Wickham, and Sabine Landau. 'Software for Generating Liability Distributions for Pedigrees Conditional on Their Observed Disease States and Covariates'. Genetic Epidemiology 34, no. 2 (1 February 2010): 159-70. doi:10.1002/gepi.20446.
          
  • Sinnwell, Jason P., Terry M. Therneau, and Daniel J. Schaid. 'The kinship2 R Package for Pedigree Data'. Human Heredity 78, no. 2 (2014): 91-93. doi:10.1159/000363105.
          
  • RStudio and Inc. (2014). shiny: Web Application Framework for R. R package version 0.10.2.2. http://CRAN.R-project.org/package=shiny.
          
  • Eddelbuettel, Dirk, and Romain, François. 'Rcpp: Seamless R and C++ Integration'. Journal of Statistical Software 40, no. 8 (2011): 1-18.
          

          
Citation

          
If you have used this website for risk calculation and wish to cite it, please use the following

          
    
          
  • Campbell, Desmond D., Yiming Li, Pak C. Sham, Stacey Cherny ...
          

          
Contacts

          
The writer and main contact regarding this work is

          
Desmond Campbell
Email: ddc123 'at' hku.hk
Tel: +852 28315156

          L10-64 Lab Block, Faculty of Medicine, 21 Sassoon Road, Pokfulam Road, Central & Western District, Hong Kong

          

          
Yiming Li
Email: liym1018 'at' hku.hk
Tel: +852 53451983

          L1-05, HKJCBIR, 5 Sassoon Road, Pokfulam, Hong Kong

          

          
Failing that, try contacting any of the other authors at the University of Hong Kong, via either the Dept of Psychiatry or the Centre for Genomic Sciences.

        

        

          

            

              
            

            

              
Download Details

              
Command Line Program

              
A command line program that does the same job as this website is available for download. This has some extra functionality over that provided by the website

              
    
              
  • It can estimate the precision of risk estimates generated
              
  • It generates a file containing 1000s of draws from the pedigree's joint liability distribution after conditioning on all disease risk relevant information
              

              
A word of warning, the installation involves installing some R packages and Rtools, so is not worth it unless you want the extra functionality.

              
Validiation Testing

              
The download also contains the testing done to validate that the command line program and the website correctly estimate risk.

              
Instructions for Installation

              
    
              
  • Download the zip file containing the program, documentation, example files, etc.
              
  • Unzip the zip file on your local machine
              
  • Read the README.html
              

            

          

        

        

          
Acknowledgements

          
Construction of this website was supported by the following.
          
EU-GEI

          
The aim of the EU-GEI project (EUropean Network of national schizophrenia networks studying Gene-Environment Interactions)
          is to identify the interactive genetic, clinical and environmental determinants, involved in the development, severity and outcome of schizophrenia.
          
The EU-GEI project is funded by the European Community's Seventh Framework Programme grant HEALTH-F2-2010-241909 (Project EU-GEI).

          
Research Grants Council, Hong Kong

          
The funding obtained from the RGC was via the General Research Fund, specifically
          
    
          
  • Project Title: Method and Software for Personal Risk Profiling of Complex Diseases
          
  • Applied for 2011/12
          
  • RGC Ref No.777511