R package:Rlmt- Overview

Rlmt: R warrper of the Linear Mixed Models Toolbox (LMT)

Contents

• OVERVIEW

• GETTING STARTED

• USAGE

OVERVIEW

​ The Linear mixed Models Toolbox (lmt) is a stand-alone single executable software for for large scale linear mixed model analysis. lmt has been used successfully for genetic evaluation data sets with »200k genotyped animals, »15m animals, »500m equations. More details about lmt can be found in website: lmt website

​ In order to further assistance the use of lmt, i developed an R package: Rlmt for interfacing with lmt. This R package is developed via R6 package which can provides an implementation of encapsulated object-oriented programming for R. And this makes the grammar of Rlmt looks like easy and elegant (i hope 😄).

​ About the usage of lmt in academic and commercial, user should follow the rules as mentioned in lmt website.

GETTING STARTED

🙊Installation

Rlmt links to R packages R6. This package should be installed before installing Rlmt. Rlmt is available by sending a request to the author (hzau_qsmei@163.com),

🙊Examples

• Ex 1. PBLUP
• Ex 2. SSGBLUP
• Ex 3. Multiple traits analysis
• Ex 4. Several special models
• Ex 5. Switch to different BLUP analysis
• Ex 6. Modify jobs

Usage

Rlmt provides several files which are saved in ~/Rlmt/extdata. We can get the path of these files by typing

system.file("extdata", package = "Rlmt") # path of provided files

Ex 1. PBLUP

library(Rlmt)
# path of example data  
example_path=system.file("extdata", package = "Rlmt") 
#construct lmt_data object
mydata=lmt_data$new(phe_file=paste0(example_path,"/data.csv"),					
                    ped_file=paste0(example_path,"/pedigree_unsorted.csv")) 

#construct  lmt_models object
mymodels=lmt_models$new(fixed=tr1~f11+f12+f13,covariate=~c1c1,random=~id)

#construct Rlmt object
mylmt=Rlmt$new(models=mymodels,data=mydata,software_path="....../lmt")

#do analysis
mylmt$do_analysis("....../test_Result") #output path

Ex 2. SSGBLUP

library(Rlmt)
# path of example data  
example_path=system.file("extdata", package = "Rlmt") 
#construct lmt_data object
mydata=lmt_data$new(phe_file=paste0(example_path,"/data.csv"),				    
                    ped_file=paste0(example_path,"/pedigree_unsorted.csv"),					
                    geno_file=paste0(example_path,"/genotypes.txt"),					
                    geno_id_file=paste0(example_path,"/genoids.csv"))

#construct  lmt_models object
mymodels=lmt_models$new(fixed=tr1~f11+f12+f13,covariate=~c1c1,random=~id)
mymodels$pars$blup_type="SS_GBLUP"
#construct Rlmt object
mylmt=Rlmt$new(models=mymodels,data=mydata,software_path="....../lmt")

#do analysis
mylmt$do_analysis("....../Rlmt/test_Result") #output path                

Ex 3. Multiple traits analysis

m2=lmt_formulas$new(fixed=tr2~f11+f12+f13,covariate=~c1c1,random=~id) #creat lmt_formulas object
mylmt$models$add_formulas(m2)
mylmt$do_analysis("....../test_Result") #multple traits analysis

m2=lmt_formulas$new(fixed=tr2~f11+f13,covariate=~c1c1+c1c2+c2c1,random=~id) #modify effect of tr2
mylmt$models$add_formulas(m2)
mylmt$do_analysis("....../Rlmt/test_Result") #output path    
mylmt$models$rm_formulas("tr2") #removes trait: tr2 in analysis
mylmt$do_analysis("....../Rlmt/test_Result") #output path  

Ex 4. Several special models

# maternal effect
m1=lmt_formulas$new(fixed=tr1~f11+f12+f13,covariate=~c1c1,random=~id+dam) 

# permanent effect
m1=lmt_formulas$new(fixed=tr1~f11+f12+f13,covariate=~c1c1,random=~id+dam+pe) 

#nested  effect
m1=lmt_formulas$new(fixed=tr1~f11+f12,covariate=~f13(c1c1),random=~id+dam+pe)

#user defined polynomial expression 
m1=lmt_formulas$new(fixed=tr1~f11+f12,covariate=~c1c1,random=~id+dam+pe,,polyno=c1c1~{x^2}+{exp(x)}) 

#3-order Legendre polynomials
m1=lmt_formulas$new(fixed=tr1~f11+f12,covariate=~c1c1,random=~id+dam+pe,,polyno=c1c1~{l1}+{l2}+{l3}) 

#random regression with 3-order Legendre polynomials
m1=lmt_formulas$new(fixed=tr1~f11+f12+f13,random=~c1c1(id)+c1c1(dam)+c1c1(pe),polyno=c1c1~{l1}+{l2}+{l3}) 

 # genetic group as extra random effect
mylmt$models$pars$gg_type="extra"

mylmt$models$add_formulas(m1)

mylmt$models$add_formulas(m1) #update effect
mylmt$do_analysis("....../test_Result") #do analysis

Ex 5. Switch to different BLUP analysis

library(Rlmt)
# path of example data  
example_path=system.file("extdata", package = "Rlmt") 
#construct lmt_data object
mydata=lmt_data$new(phe_file=paste0(example_path,"/data.csv"),				    
                    ped_file=paste0(example_path,"/pedigree_unsorted.csv"),					
                    geno_file=paste0(example_path,"/genotypes.txt"),					
                    geno_id_file=paste0(example_path,"/genoids.csv"))

#construct  lmt_models object
mymodels=lmt_models$new(fixed=tr1~f11+f12+f13,covariate=~c1c1,random=~id)

mylmt$models$pars$meta_gamma=0.5 #Metafounder

mymodels$pars$blup_type="SS_TBLUP" #SS_TBLUP

mymodels$pars$blup_type="SS_SNPBLUP" #SS_SNPBLUP

mymodels$pars$blup_type="External_BLUP"   #External_BLUP
mylmt$data$grm$file="test_external.bin"   #External_BLUP

#ssGBLUP with two genomic factors
mydata=lmt_data$new(phe_file=paste0(example_path,"/data.csv"),                  
                    ped_file=paste0(example_path,"/pedigree_unsorted.csv"),					
                    geno_file=paste0(example_path,c("/genotypes1.txt","/genotypes2.txt")),			
                    geno_id_file=paste0(example_path,c("/genoids1.txt","/genoids2.txt")))
mymodels=lmt_models$new(fixed=tr1~f11+f12+f13,covariate=~c1c1,random=~id1+id2)                    
     

mylmt$do_analysis("/usr/home/qgg/qumei/Rlmt/test_Result") #do analysis

Ex 6. Modify jobs

#sampler 
m_jobs=lmt_jobs$new(type="sample")
mymodels$pars$jobs=m_jobs
m_jobs$samplers$type=c("singlepass","blocked")
mylmt$do_analysis("....../test_Result") #output path

		
#solver 
m_jobs=mymodels$pars$jobs
m_jobs$type="mcemreml"
m_jobs$conv=-9.21034
m_jobs$rounds=300
mylmt$do_analysis("....../test_Result") #output path

#pevsolve
m_jobs=lmt_jobs$new(type="pevsolve")
mymodels$pars$jobs=m_jobs
mylmt$do_analysis("....../test_Result") #output path