https://github.com/gregor-mendel-institute/mtmm

A mixed-model approach for genome-wide association studies of correlated traits in structured populations
https://github.com/gregor-mendel-institute/mtmm

Last synced: 7 days ago
JSON representation

A mixed-model approach for genome-wide association studies of correlated traits in structured populations

• Host: GitHub
• URL: https://github.com/gregor-mendel-institute/mtmm
• Owner: Gregor-Mendel-Institute
• Created: 2015-03-10T14:49:42.000Z (over 9 years ago)
• Default Branch: master
• Last Pushed: 2018-06-22T20:00:59.000Z (over 6 years ago)
• Last Synced: 2024-03-26T21:06:39.284Z (8 months ago)
• Language: R
• Homepage: http://www.nature.com/ng/journal/v44/n9/full/ng.2376.html
• Size: 54.9 MB
• Stars: 7
• Watchers: 12
• Forks: 8
• Open Issues: 2
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# MTMM - A mixed-model approach for genome-wide association studies of correlated traits in structured populations

##  Introduction

The MTMM function as published in [Nature Genetics](http://www.nature.com/ng/journal/v44/n9/full/ng.2376.html) currently don't support estimates on missing data and replicates. 

This is work in progress and will be accordingly updated here.

For questions and comments feel free to contact me: [email protected]

 

## How to use

```R

# Load libraries and source needed functions

# The AsREML package needs a valid license that can be obtained at  http://www.vsni.co.uk/software/asreml

library(lattice)

library(asreml)

# msm and nadiv librarys are used to estimate SE of the correlation estimates, only used if run=FALSE 

#library(msm)

#library(nadiv)

source('mtmm_function.r')

source('emma.r')

# load your data (Phenotype(Y),Genotype(X) and Kinship(K))

# note you can calculate K using the emma package K<-emma.kinship(t(X)), make sure to set colnames(K)=rownames(K)=rownames(X)

# alternativley load the sample data

load('data/MTMM_SAMPLE_DATA.Rdata')

# different options include method(default or errorcorrelation, include.single.analysis, calculate.effect.size (if TRUE, #analysis is more time consuming) default for X is binary coding of 0 and 1, if your data are code  0,1 and 2 use #gen.data='heterozygot',  run=FALSE will not perform the GWAS, but only output the correlation estimates (fast)

mtmm(Y,X,K,method='default',include.single.analysis=T,calculate.effect.size=T,gen.data='binary',exclude=T,run=T)

# To only perform a Variance Coponent Analysis use the mtmm_estimate.r script with the flag only.vca=T set

VCA<-mtmm_estimates(Y,K=K,only.vca=T)

# the function outputs a list called results  ($phenotype ,$pvals, $statistics, $kinship)

output<-results$pvals

# manhattan plots

# default plots for include.single.analysis=T

par(mfrow=c(5,1),mar=c(3, 4, 1, 4))

plot_gwas(output,h=8)

plot_gwas(output,h=9)

plot_gwas(output,h=10)

plot_gwas(output,h=11)

plot_gwas(output,h=12)

#qq plots

par(mfrow=c(1,1),mar=c(3, 4, 1, 4))

qq_plot_all(output)

```

## Poster

* [Complextraits 2012](posters/poster_complextraits_2012_AK.pdf)

* [ICAR 2012](posters/poster_ICAR_2012_AK.pptx)