Genomic Prediction Analysis - Stage II of II: Cross-validation Round 2
wolfemd
2019-7-29
Last updated: 2019-11-21
Checks: 7 0
Knit directory: IITA_2019GS/
This reproducible R Markdown analysis was created with workflowr (version 1.5.0.9000). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.
Great! Since the R Markdown file has been committed to the Git repository, you know the exact version of the code that produced these results.
Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.
The command set.seed(20191121) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.
Great job! Recording the operating system, R version, and package versions is critical for reproducibility.
Nice! There were no cached chunks for this analysis, so you can be confident that you successfully produced the results during this run.
Great job! Using relative paths to the files within your workflowr project makes it easier to run your code on other machines.
Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility. The version displayed above was the version of the Git repository at the time these results were generated.
Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated:
Ignored files:
Ignored: .DS_Store
Ignored: .Rproj.user/
Ignored: data/.DS_Store
Ignored: output/.DS_Store
Untracked files:
Untracked: data/GBSdataMasterList_31818.csv
Untracked: data/IITA_CleanedTrialData_72519.rds
Untracked: data/IITA_ExptDesignsDetected_72619.rds
Untracked: data/IITA_GBStoPhenoMaster_33018.csv
Untracked: data/NRCRI_GBStoPhenoMaster_40318.csv
Untracked: data/iita_blupsForCrossVal_72619.rds
Untracked: data/iita_blupsForCrossVal_outliersRemoved_73019.rds
Untracked: output/CrossVal_72719/
Untracked: output/CrossVal_73019/
Untracked: output/GEBV_BRNHT1_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_BRNHT1_OutliersEqualTRUE_IITA_73119.rds
Untracked: output/GEBV_DM_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_DM_OutliersEqualTRUE_IITA_73119.rds
Untracked: output/GEBV_DM_white_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_DM_white_OutliersEqualTRUE_IITA_73119.rds
Untracked: output/GEBV_DM_yellow_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_DM_yellow_OutliersEqualTRUE_IITA_73119.rds
Untracked: output/GEBV_HI_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_HI_OutliersEqualTRUE_IITA_73119.rds
Untracked: output/GEBV_IITA_OutliersRemovedFALSE_73119.csv
Untracked: output/GEBV_IITA_OutliersRemovedTRUE_73119.csv
Untracked: output/GEBV_MCMDS_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_MCMDS_OutliersEqualTRUE_IITA_73119.rds
Untracked: output/GEBV_PLTHT_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_PLTHT_OutliersEqualTRUE_IITA_73119.rds
Untracked: output/GEBV_RTSZ_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_RTSZ_OutliersEqualTRUE_IITA_73119.rds
Untracked: output/GEBV_TCHART_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_TCHART_OutliersEqualTRUE_IITA_73119.rds
Untracked: output/GEBV_logFYLD_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logFYLD_OutliersEqualTRUE_IITA_73119.rds
Untracked: output/GEBV_logFYLD_lowCMD_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logFYLD_nohav5_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logFYLD_propHAVpt5_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logFYLD_propHAVpt75_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logRTNO_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logRTNO_OutliersEqualTRUE_IITA_73119.rds
Untracked: output/GEBV_logRTNO_lowCMD_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logRTNO_nohav5_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logRTNO_propHAVpt5_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logRTNO_propHAVpt75_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logTOPYLD_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logTOPYLD_OutliersEqualTRUE_IITA_73119.rds
Untracked: output/GEBV_logTOPYLD_lowCMD_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logTOPYLD_nohav5_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logTOPYLD_propHAVpt5_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/GEBV_logTOPYLD_propHAVpt75_OutliersEqualFALSE_IITA_73119.rds
Untracked: output/PCscores_IITA_SamplesForGP_73119.rds
Untracked: output/sketch_of_crossVal_scheme.png
Untracked: workflowr_log.R
Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.
These are the previous versions of the R Markdown and HTML files. If you’ve configured a remote Git repository (see ?wflow_git_remote), click on the hyperlinks in the table below to view them.
| File | Version | Author | Date | Message |
|---|---|---|---|---|
| html | a869b9e | wolfemd | 2019-11-21 | Build site. |
| Rmd | bfffb51 | wolfemd | 2019-11-21 | Publish the first set of analyses and files for IITA 2019 GS, |
Objective
This time with the outliers-removed BLUPs. Based on results in round 1, did not continue with some of the traits.
Set-up training data
rm(list=ls()); gc()
library(tidyverse); library(magrittr);
blups<-readRDS(file="data/iita_blupsForCrossVal_outliersRemoved_73019.rds")
K<-readRDS(file=paste0("/workdir/IITA_2019GS/Kinship_IITA_TrainingPop_72619.rds"))
blups %<>%
rename(trainingData=blups) %>%
mutate(trainingData=map(trainingData,~filter(.,GID %in% rownames(K))),)
tms13f<-rownames(K) %>% grep("TMS13F|2013_",.,value = T); length(tms13f) # 2395
tms14f<-rownames(K) %>% grep("TMS14F",.,value = T); length(tms14f) # 2171
tms15f<-rownames(K) %>% grep("TMS15F",.,value = T); length(tms15f) # 835
gg<-setdiff(rownames(K),c(tms13f,tms14f,tms15f)); length(gg) # 1228 (not strictly gg)
blups %<>%
mutate(seed_of_seeds=1:n(),
seeds=map(seed_of_seeds,function(seed_of_seeds,reps=5){
set.seed(seed_of_seeds);
outSeeds<-sample(1:1000,size = reps,replace = F);
return(outSeeds) }))
blups %<>%
select(-varcomp); gc()Cross-validation function
# trainingData<-blups$trainingData[[1]]; seeds<-blups$seeds[[1]]; nfolds<-5; reps<-5;
crossValidateFunc<-function(Trait,trainingData,seeds,nfolds=5,reps=5,ncores=50,...){
trntstdata<-trainingData %>%
filter(GID %in% rownames(K))
K1<-K[rownames(K) %in% trntstdata$GID,
rownames(K) %in% trntstdata$GID]
# rm(K,trainingData); gc()
# seed<-seeds[[1]]
# Nfolds=nfolds
makeFolds<-function(Nfolds=nfolds,seed){
genotypes<-rownames(K1)
set.seed(seed)
seed_per_group<-sample(1:10000,size = 4,replace = FALSE)
set.seed(seed_per_group[1])
FoldsThisRep_tms15<-tibble(CLONE=genotypes[genotypes %in% tms15f],
Group="TMS15F") %>%
mutate(Folds=sample(1:Nfolds,nrow(.),replace=T)) %>%
arrange(Folds) %>%
group_by(Group,Folds) %>%
nest(.key = Test)
set.seed(seed_per_group[2])
FoldsThisRep_tms14<-tibble(CLONE=genotypes[genotypes %in% tms14f],
Group="TMS14F") %>%
mutate(Folds=sample(1:Nfolds,nrow(.),replace=T)) %>%
arrange(Folds) %>%
group_by(Group,Folds) %>%
nest(.key = Test)
set.seed(seed_per_group[3])
FoldsThisRep_tms13<-tibble(CLONE=genotypes[genotypes %in% tms13f],
Group="TMS13F") %>%
mutate(Folds=sample(1:Nfolds,nrow(.),replace=T)) %>%
arrange(Folds) %>%
group_by(Group,Folds) %>%
nest(.key = Test)
set.seed(seed_per_group[4])
FoldsThisRep_gg<-tibble(CLONE=genotypes[genotypes %in% gg],
Group="GGetc") %>%
mutate(Folds=sample(1:Nfolds,nrow(.),replace=T)) %>%
arrange(Folds) %>%
group_by(Group,Folds) %>%
nest(.key = Test)
FoldsThisRep<-bind_rows(FoldsThisRep_tms15,FoldsThisRep_tms14) %>%
bind_rows(FoldsThisRep_tms13) %>%
bind_rows(FoldsThisRep_gg) %>%
mutate(Test=map(Test,~.$CLONE),
Train=map(Test,~genotypes[!genotypes %in% .]))
return(FoldsThisRep) }
crossval<-tibble(Rep=1:reps,seed=unlist(seeds)[1:reps]) %>%
mutate(Folds=map2(Rep,seed,~makeFolds(Nfolds=nfolds,seed=.y))) %>%
unnest()
#Test<-crossval$Test[[1]]; Train<-crossval$Train[[1]]
crossValidate<-function(Test,Train){
train<-Train
test<-Test
trainingdata<-trntstdata %>%
filter(GID %in% train) %>%
mutate(GID=factor(GID,levels=rownames(K1)))
require(sommer)
proctime<-proc.time()
fit <- mmer(fixed = drgBLUP ~1,
random = ~vs(GID,Gu=K1),
weights = WT,
data=trainingdata)
proc.time()-proctime
x<-fit$U$`u:GID`$drgBLUP
gebvs<-tibble(GID=names(x),
GEBV=as.numeric(x))
accuracy<-gebvs %>%
filter(GID %in% test) %>%
left_join(
trntstdata %>%
dplyr::select(GID,BLUP) %>%
filter(GID %in% test)) %$%
cor(GEBV,BLUP, use='complete.obs')
return(accuracy)
}
require(furrr)
options(mc.cores=ncores)
plan(multiprocess)
crossval<-crossval %>%
mutate(accuracy=future_map2(Test,Train,~crossValidate(Test=.x,Train=.y)))
saveRDS(crossval,file=paste0("/workdir/IITA_2019GS/CrossVal_73019/",
"CrossVal_",Trait,"_OutliersRemoved_73019.rds"))
rm(list=ls()); gc()
}Run CV on two servers
cbsulm14 (112)
Results
used (Mb) gc trigger (Mb) limit (Mb) max used (Mb)
Ncells 605197 32.4 1259562 67.3 NA 886271 47.4
Vcells 1158306 8.9 8388608 64.0 102400 1958857 15.0library(tidyverse); library(magrittr); library(cowplot);
cvNoOutliers<-tibble(Files=list.files("output/CrossVal_73019/")) %>%
mutate(Trait=gsub("CrossVal_","",Files),
Trait=gsub("_OutliersRemoved_73019.rds","",Trait),
Dataset="OutliersRemoved") %>%
mutate(cvResults=map(Files,~readRDS(paste0("output/CrossVal_73019/",.)))) %>%
dplyr::select(-Files)
cvWithOutliers<-tibble(Files=list.files("output/CrossVal_72719/")) %>%
filter(grepl("HistoricalDataIncluded|BRNHT1|PLTHT",Files)) %>%
mutate(Trait=gsub("CrossVal_","",Files),
Trait=gsub("_2013toPresent_72719.rds","",Trait),
Trait=gsub("_HistoricalDataIncluded_72719.rds","",Trait),
Dataset="NoOutlierRemoval") %>%
filter(Trait %in% cvNoOutliers$Trait) %>%
mutate(cvResults=map(Files,~readRDS(paste0("output/CrossVal_72719/",.)))) %>%
dplyr::select(-Files)
cv<-bind_rows(cvNoOutliers,
cvWithOutliers)
cv %<>%
unnest(cols = cvResults) %>%
mutate(Ntrain=map_dbl(Train,length),
Ntest=map_dbl(Test,length)) %>%
select(-Test,-Train) %>%
unnest(cols = accuracy)Figure 1
I did an additional cross-validation, using BLUPs produced after two rounds of model-fitting, followed-by outlier removal. I defined outliers as observations with abs(studentized residuals)>3.3. Overall, the improvement is not consistent or large, but I’d probably trend towards using the data with outliers removed.
By genetic group
library(viridis)
cv %>%
ggplot(.,aes(x=Trait,y=accuracy,fill=Dataset)) +
geom_boxplot() +
facet_grid(.~Group,space='free_x',scale='free_x') +
geom_hline(yintercept = 0,color='darkred',size=1.25) +
theme_bw() +
theme(axis.text.x = element_text(angle=90,face='bold',size=14)) +
scale_fill_viridis_d()
| Version | Author | Date |
|---|---|---|
| a869b9e | wolfemd | 2019-11-21 |
Figure 2
overall
library(viridis)
cv %>%
ggplot(.,aes(x=Trait,y=accuracy,fill=Dataset)) +
geom_boxplot() +
geom_hline(yintercept = 0,color='darkred',size=1.25) +
theme_bw() +
theme(axis.text.x = element_text(angle=90,face='bold',size=14)) +
scale_fill_viridis_d()
| Version | Author | Date |
|---|---|---|
| a869b9e | wolfemd | 2019-11-21 |
R version 3.6.1 (2019-07-05)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS Mojave 10.14.6
Matrix products: default
BLAS: /Library/Frameworks/R.framework/Versions/3.6/Resources/lib/libRblas.0.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.6/Resources/lib/libRlapack.dylib
locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] viridis_0.5.1 viridisLite_0.3.0 cowplot_1.0.0 magrittr_1.5
[5] forcats_0.4.0 stringr_1.4.0 dplyr_0.8.3 purrr_0.3.3
[9] readr_1.3.1 tidyr_1.0.0 tibble_2.1.3 ggplot2_3.2.1
[13] tidyverse_1.2.1
loaded via a namespace (and not attached):
[1] tidyselect_0.2.5 xfun_0.11 reshape2_1.4.3
[4] haven_2.2.0 lattice_0.20-38 colorspace_1.4-1
[7] vctrs_0.2.0 generics_0.0.2 htmltools_0.4.0
[10] yaml_2.2.0 rlang_0.4.1 later_1.0.0
[13] pillar_1.4.2 withr_2.1.2 glue_1.3.1
[16] modelr_0.1.5 readxl_1.3.1 plyr_1.8.4
[19] lifecycle_0.1.0 munsell_0.5.0 gtable_0.3.0
[22] workflowr_1.5.0.9000 cellranger_1.1.0 rvest_0.3.5
[25] evaluate_0.14 labeling_0.3 knitr_1.26
[28] httpuv_1.5.2 broom_0.5.2 Rcpp_1.0.3
[31] promises_1.1.0 backports_1.1.5 scales_1.1.0
[34] jsonlite_1.6 farver_2.0.1 fs_1.3.1
[37] gridExtra_2.3 hms_0.5.2 digest_0.6.22
[40] stringi_1.4.3 grid_3.6.1 rprojroot_1.3-2
[43] cli_1.1.0 tools_3.6.1 lazyeval_0.2.2
[46] crayon_1.3.4 whisker_0.4 pkgconfig_2.0.3
[49] zeallot_0.1.0 xml2_1.2.2 lubridate_1.7.4
[52] assertthat_0.2.1 rmarkdown_1.17 httr_1.4.1
[55] rstudioapi_0.10 R6_2.4.1 nlme_3.1-142
[58] git2r_0.26.1 compiler_3.6.1