Last updated: 2020-10-16
Checks: 7 0
Knit directory: NRCRI_2020GS/
This reproducible R Markdown analysis was created with workflowr (version 1.6.2). 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(20200421) 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 results in this page were generated with repository version c5c0337. See the Past versions tab to see a history of the changes made to the R Markdown and HTML files.
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: .Rhistory
Ignored: .Rproj.user/
Ignored: analysis/.DS_Store
Ignored: data/.DS_Store
Ignored: output/.DS_Store
Untracked files:
Untracked: CrossValidationFunction_PlusNRCRI_2020GSnotes.gslides
Untracked: data/DatabaseDownload_2020Oct13/
Untracked: data/chr1_RefPanelAndGSprogeny_ReadyForGP_72719.fam
Untracked: output/Kinship_AA_NRCRI_2020April27.rds
Untracked: output/Kinship_AD_NRCRI_2020April27.rds
Untracked: output/Kinship_AD_NRCRI_2020Oct15.rds
Untracked: output/Kinship_A_NRCRI_2020April27.rds
Untracked: output/Kinship_A_NRCRI_2020Oct15.rds
Untracked: output/Kinship_DD_NRCRI_2020April27.rds
Untracked: output/Kinship_D_NRCRI_2020April27.rds
Untracked: output/Kinship_D_NRCRI_2020Oct15.rds
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 repository in which changes were made to the R Markdown (analysis/08-CrossValidation.Rmd) and HTML (docs/08-CrossValidation.html) files. If you’ve configured a remote Git repository (see ?wflow_git_remote), click on the hyperlinks in the table below to view the files as they were in that past version.
| File | Version | Author | Date | Message |
|---|---|---|---|---|
| html | 12bd17b | wolfemd | 2020-10-16 | Build site. |
| Rmd | 977f389 | wolfemd | 2020-10-16 | Publish NRCRI imputations for 2020 (DCas20_5510 and DCas20_5440) plus a |
Get BLUPs combining all trial data: Combine data from all trait-trials to get BLUPs for downstream genomic prediction.
Current Step:
Check prediction accuracy: Evaluate prediction accuracy with cross-validation.
5-fold cross-validation. Replicate 5-times.
2 genomic models:
The data for the next step can be found on the cassavabase FTP server here.
Can be loaded directly to R from FTP.
NOTICE: You need enough RAM and a stable network connection. I do the next steps, including cross-validation on a server with plenty of RAM and a good, stable network connection, rather than on my personal computer (a laptop with 16 GB RAM).
The outputs (kinship matrices and filtered snp dosages) of the steps below, which are too large for GitHub, can be found on the cassavabase FTP server here.
# activate multithread OpenBLAS for fast compute of SigmaM (genotypic var-covar matrix)
export OMP_NUM_THREADS=56library(tidyverse); library(magrittr);
snps<-readRDS(file=url(paste0("ftp://ftp.cassavabase.org/marnin_datasets/NGC_BigData/",
"DosageMatrix_RefPanelAndGSprogeny_ReadyForGP_73019.rds")))
snps5510<-readRDS(here::here("output","DosageMatrix_DCas20_5510_WA_REFimputedAndFiltered.rds"))
snps5440<-readRDS(here::here("output","DosageMatrix_DCas20_5440_WA_REFimputedAndFiltered.rds"))
rownames(snps5510) # NR19C3aF1P0002....
rownames(snps5440) # NR18S1P0082...
snps2keep<-colnames(snps) %>%
.[. %in% colnames(snps5510)] %>%
.[. %in% colnames(snps5440)]
snps<-rbind(snps[,snps2keep],
snps5510[,snps2keep]) %>%
rbind(.,snps5440[,snps2keep])
gc()
#rm(list=(ls() %>% grep("snps",.,value = T, invert = T)))blups_nrcri<-readRDS(file=here::here("output","nrcri_blupsForModelTraining_twostage_asreml_2020Oct13.rds"))
# copy or download IITA BLUPs from GitHub to data/ directory
# https://wolfemd.github.io/IITA_2020GS/output/iita_blupsForModelTraining_twostage_asreml.rds
blups_iita<-readRDS(file=here::here("data","iita_blupsForModelTraining_twostage_asreml.rds"))
blups_nrcri %<>%
select(Trait,blups) %>%
unnest(blups) %>%
select(-`std error`) %>%
filter(GID %in% rownames(snps))
table(unique(blups_nrcri$GID) %in% rownames(snps)) # 2751
blups_iita %<>%
select(Trait,blups) %>%
unnest(blups) %>%
select(-`std error`) %>%
filter(GID %in% rownames(snps),
!grepl("TMS13F|TMS14F|TMS15F|2013_|TMS16F|TMS17F|TMS18F",GID))
table(unique(blups_iita$GID) %in% rownames(snps)) # 1234Going to use my own kinship function.
Make the kinships.
Below e.g. A*A makes a matrix that approximates additive-by-additive epistasis relationships.
A<-kinship(snps,type="add")
D<-kinship(snps,type="dom")
AD<-A*D
saveRDS(snps,file=here::here("output","DosageMatrix_NRCRI_2020Oct15.rds"))
saveRDS(A,file=here::here("output","Kinship_A_NRCRI_2020Oct15.rds"))
saveRDS(D,file=here::here("output","Kinship_D_NRCRI_2020Oct15.rds"))
saveRDS(AD,file=here::here("output","Kinship_AD_NRCRI_2020Oct15.rds"))
#rm(snps); gc()NOTICE: The outputs (kinship matrices and filtered snp dosages) of the steps below, which are too large for GitHub, can be found on the cassavabase FTP server here.
rm(list=ls())
library(tidyverse); library(magrittr);
source(here::here("code","gsFunctions.R"))
blups_nrcri<-readRDS(file=here::here("output","nrcri_blupsForModelTraining_twostage_asreml_2020Oct13.rds"))
blups_iita<-readRDS(file=here::here("data","iita_blupsForModelTraining_twostage_asreml.rds"))
A<-readRDS(file=here::here("output","Kinship_A_NRCRI_2020Oct15.rds"))
blups_nrcri %<>%
select(Trait,blups) %>%
unnest(blups) %>%
select(-`std error`) %>%
filter(GID %in% rownames(A))
blups_iita %<>%
select(Trait,blups) %>%
unnest(blups) %>%
select(-`std error`) %>%
filter(GID %in% rownames(A),
!grepl("TMS13F|TMS14F|TMS15F|2013_|TMS16F|TMS17F|TMS18F",GID))
# Set-up a grouping variable for:
## nrTP, C1a, C1b and C2a.
## Nest by Trait.
c1a<-blups_nrcri$GID %>%
unique %>%
grep("c1a",.,value = T,ignore.case = T) %>%
union(.,blups_nrcri$GID %>% unique %>%
grep("^F",.,value = T,ignore.case = T) %>%
grep("c1b",.,value = T,ignore.case = T,invert = T))
c1b<-blups_nrcri$GID %>% unique %>% grep("c1b",.,value = T,ignore.case = T)
c2a<-blups_nrcri$GID %>% unique %>%
grep("C2a",.,value = T,ignore.case = T) %>%
grep("NR17",.,value = T,ignore.case = T)
c2b<-blups_nrcri$GID %>% unique %>%
grep("C2b",.,value = T,ignore.case = T) %>%
.[!. %in% c(c1a,c1b,c2a)]
nrTP<-setdiff(unique(blups_nrcri$GID),unique(c(c1a,c1b,c2a,c2b)))
cv2do<-blups_nrcri %>%
mutate(Group=case_when(GID %in% nrTP ~ "nrTP",
GID %in% c1a ~ "C1a",
GID %in% c1b ~ "C1b",
GID %in% c2a ~ "C2a",
GID %in% c2b ~ "C2b")) %>%
nest(TrainTestData=-Trait) %>%
left_join(blups_iita %>%
nest(augmentTP=-Trait))
cv2do %>% rmarkdown::paged_table()The function below runCrossVal() function implements nfold cross-validation. Specifically, for each of nrepeats it splits the data into nfolds sets according to gid. So if nfolds=5 then the the clones will be divided into 5 groups and 5 predictions will be made. In each prediction, 4/5 of the clones will be used to predict the remaining 1/5. Accuracy of the model is measured as the correlation between the BLUPs (adj. mean for each CLONE) in the test set and the GEBV (the prediction made of each clone when it was in the test set).
Below, 4 chunks of “5 reps x 5-fold” cross-validation are run on 1 large memory Cornell CBSU machine each (e.g. cbsulm16; 112 cores, 512 GB RAM).
starttime<-proc.time()[3]
cv_A_nrOnly<-cv2do %>%
mutate(CVresults=map(TrainTestData,~runCrossVal(TrainTestData=.,
modelType="A",
grms=list(A=A),
byGroup=TRUE,augmentTP=NULL,
nrepeats=5,nfolds=5,ncores=4,gid="GID")))
runtime<-proc.time()[3]-starttime; runtime
cv_A_nrOnly %<>% mutate(Dataset="NRalone",modelType="A") %>% dplyr::select(-TrainTestData,-augmentTP)
saveRDS(cv_A_nrOnly,file=here::here("output","cvresults_A_nrOnly_2020Oct15.rds"))For this one, try with ncores=1 instead of ncores=10.
starttime<-proc.time()[3]
cv_A_iitaAugmented<-cv2do %>%
mutate(isnullAugment=map_lgl(augmentTP,~is.null(.))) %>%
filter(!isnullAugment) %>%
select(-isnullAugment) %>%
mutate(CVresults=map2(TrainTestData,augmentTP,~runCrossVal(TrainTestData=.x,
modelType="A",
grms=list(A=A),
byGroup=TRUE,augmentTP=.y,
nrepeats=5,nfolds=5,ncores=1,gid="GID")))
runtime<-proc.time()[3]-starttime; runtime
cv_A_iitaAugmented %<>% mutate(Dataset="IITAaugmented",modelType="A") %>% dplyr::select(-TrainTestData,-augmentTP)
saveRDS(cv_A_iitaAugmented,file=here::here("output","cvresults_A_iitaAugmented_2020Oct15.rds"))options(future.globals.maxSize= 3000*1024^2)
D<-readRDS(file=here::here("output","Kinship_D_NRCRI_2020Oct15.rds"))
AD<-readRDS(file=here::here("output","Kinship_AD_NRCRI_2020Oct15.rds"))
starttime<-proc.time()[3]
cv_ADE_nrOnly<-cv2do %>%
mutate(CVresults=map(TrainTestData,~runCrossVal(TrainTestData=.,
modelType="ADE",
grms=list(A=A,D=D,AD=AD),
byGroup=TRUE,augmentTP=NULL,
nrepeats=5,nfolds=5,ncores=10,gid="GID")))
cv_ADE_nrOnly %<>% mutate(Dataset="NRalone",modelType="ADE") %>% dplyr::select(-TrainTestData,-augmentTP)
saveRDS(cv_ADE_nrOnly,file=here::here("output","cvresults_ADE_nrOnly.rds"))
runtime<-proc.time()[3]-starttime; runtimeoptions(future.globals.maxSize= 3000*1024^2)
D<-readRDS(file=here::here("output","Kinship_D_NRCRI_2020Oct15.rds"))
AD<-readRDS(file=here::here("output","Kinship_AD_NRCRI_2020Oct15.rds"))
starttime<-proc.time()[3]
cv_ADE_iitaAugmented<-cv2do %>%
mutate(isnullAugment=map_lgl(augmentTP,~is.null(.))) %>%
filter(!isnullAugment) %>%
dplyr::select(-isnullAugment) %>%
mutate(CVresults=map2(TrainTestData,augmentTP,~runCrossVal(TrainTestData=.x,
modelType="ADE",
grms=list(A=A,D=D,AD=AD),
byGroup=TRUE,augmentTP=.y,
nrepeats=5,nfolds=5,ncores=10,gid="GID")))
cv_ADE_iitaAugmented %<>% mutate(Dataset="IITAaugmented",modelType="ADE") %>% dplyr::select(-TrainTestData,-augmentTP)
saveRDS(cv_ADE_iitaAugmented,file=here::here("output","cvresults_ADE_iitaAugmented_2020Oct15.rds"))
runtime<-proc.time()[3]-starttime; runtimeoptions(future.globals.maxSize= 3000*1024^2)
D<-readRDS(file=here::here("output","Kinship_D_NRCRI_2020Oct15.rds"))
#AD<-readRDS(file=here::here("output","Kinship_AD_NRCRI_2020Oct15.rds"))
starttime<-proc.time()[3]
cv_AD_iitaAugmented<-cv2do %>%
mutate(isnullAugment=map_lgl(augmentTP,~is.null(.))) %>%
filter(!isnullAugment) %>%
dplyr::select(-isnullAugment) %>%
mutate(CVresults=map2(TrainTestData,augmentTP,~runCrossVal(TrainTestData=.x,
modelType="AD",
grms=list(A=A,D=D),
byGroup=TRUE,augmentTP=.y,
nrepeats=5,nfolds=5,ncores=4,gid="GID")))
cv_AD_iitaAugmented %<>% mutate(Dataset="IITAaugmented",modelType="AD") %>% dplyr::select(-TrainTestData,-augmentTP)
saveRDS(cv_AD_iitaAugmented,file=here::here("output","cvresults_AD_iitaAugmented_2020Oct15.rds"))
runtime<-proc.time()[3]-starttime; runtimeSee Results
R version 4.0.2 (2020-06-22)
Platform: x86_64-apple-darwin17.0 (64-bit)
Running under: macOS Mojave 10.14.6
Matrix products: default
BLAS: /Library/Frameworks/R.framework/Versions/4.0/Resources/lib/libRblas.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/4.0/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] magrittr_1.5 forcats_0.5.0 stringr_1.4.0 dplyr_1.0.2
[5] purrr_0.3.4 readr_1.4.0 tidyr_1.1.2 tibble_3.0.4
[9] ggplot2_3.3.2 tidyverse_1.3.0 workflowr_1.6.2
loaded via a namespace (and not attached):
[1] tidyselect_1.1.0 xfun_0.18 haven_2.3.1 colorspace_1.4-1
[5] vctrs_0.3.4 generics_0.0.2 htmltools_0.5.0 yaml_2.2.1
[9] blob_1.2.1 rlang_0.4.8 later_1.1.0.1 pillar_1.4.6
[13] withr_2.3.0 glue_1.4.2 DBI_1.1.0 dbplyr_1.4.4
[17] modelr_0.1.8 readxl_1.3.1 lifecycle_0.2.0 munsell_0.5.0
[21] gtable_0.3.0 cellranger_1.1.0 rvest_0.3.6 evaluate_0.14
[25] knitr_1.30 ps_1.4.0 httpuv_1.5.4 fansi_0.4.1
[29] broom_0.7.1 Rcpp_1.0.5 promises_1.1.1 backports_1.1.10
[33] scales_1.1.1 jsonlite_1.7.1 fs_1.5.0 hms_0.5.3
[37] digest_0.6.25 stringi_1.5.3 rprojroot_1.3-2 grid_4.0.2
[41] here_0.1 cli_2.1.0 tools_4.0.2 crayon_1.3.4
[45] whisker_0.4 pkgconfig_2.0.3 ellipsis_0.3.1 xml2_1.3.2
[49] reprex_0.3.0 lubridate_1.7.9 rstudioapi_0.11 assertthat_0.2.1
[53] rmarkdown_2.4 httr_1.4.2 R6_2.4.1 git2r_0.27.1
[57] compiler_4.0.2