R/BLUP_modelling.r

   1 rm(list=ls())
   2 library(lme4)
   3
   4 # Define models
   5
   6 #0# fixed effect trait, random effect (genotype)
   7 data_i <- data.frame(phenodata_sub=phenodata[,i],genotype = genotype[,1])
   8 fmer <- lmer(phenodata~1|genotype, data = data_i, na.action = na.omit)
   9
  10 #1# fixed effect: year, random effect (genotype)
  11 fmer <- lmer(phenodata[,i]~year + (1|genotype), data = phenodata)
  12 fmer <- lmer(v~1|w, data = data_i, na.action = na.omit)
  13
  14 #2# fixed effect: block, random effect (genotype)
  15 fmer <- lmer(phenodata[,i]~block + (1|genotype), data = phenodata)
  16
  17 #3# fixed effects: year and block (additive), random effect (genotype) and covariate response identical for all
  18 fmeria  <- lmer(phenodata[,i]~year + block + (1|genotype), data = phenodata)
  19
  20 #4# fixed effects: (year and block (additive),  random effect (genotype),  interaction (genotype:year) and covariate response identical for all
  21 fmerib  <- lmer(phenodata[,i]~year + block + (1|genotype) + (1|genotype:year), data = phenodata)
  22
  23 #5# fixed effects: (year and block (additive), , hierarchical random effect (genotype),(response to genotype covariate based on year):
  24 fmert  <- lmer(phenodata[,i]~year + block + (-1+year|genotype), data = phenodata)
  25
  26 # Model comparison (most complex vs less complex)
  27 #genotype effect on year+block:
  28 model(2) versus model(0)
  29
  30 #year effect on year+block:
  31 model(1) versus model(0)
  32
  33 #block effect on genotype:
  34 model(2) versus model(1)
  35
  36 #genotype:year interaction on genotype:
  37 model(3) versus model(2)
  38
  39 #year + block hierachical effect on genotype:
  40 model(5) versus model(2)