Power vs. FDR plots for paper
Yuxin Zou and Peter Carbonetto
Last updated: 2022-06-20
Checks: 7 0
Knit directory: dsc_susierss/
This reproducible R Markdown analysis was created with workflowr (version 1.7.0). 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(20210220) 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 f36f73e. 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: .Rhistory
Ignored: .Rproj.user/
Untracked files:
Untracked: cs_plots.eps
Untracked: pip_power_vs_fdr_insample.eps
Untracked: pip_power_vs_fdr_outofsample.eps
Untracked: pip_power_vs_fdr_pve.eps
Untracked: pip_power_vs_fdr_refine.eps
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/power_vs_fdr.Rmd) and HTML
(docs/power_vs_fdr.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 |
|---|---|---|---|---|
| Rmd | f36f73e | Yuxin Zou | 2022-06-20 | wflow_publish("analysis/power_vs_fdr.Rmd") |
| html | 02c5d23 | Yuxin Zou | 2022-06-20 | Build site. |
| Rmd | 94f6317 | zouyuxin | 2022-06-20 | add code generating plots |
| html | 45fa987 | Peter Carbonetto | 2022-01-26 | Added grid line to in-sample-ld power vs. fdr plot. |
| Rmd | cb03902 | Peter Carbonetto | 2022-01-26 | workflowr::wflow_publish("analysis/power_vs_fdr.Rmd", verbose = TRUE) |
| html | a03de31 | Peter Carbonetto | 2022-01-26 | Build site. |
| Rmd | 339f6ec | Peter Carbonetto | 2022-01-26 | workflowr::wflow_publish("analysis/power_vs_fdr.Rmd") |
| Rmd | ede4be4 | Peter Carbonetto | 2022-01-25 | A few small edits to the power_vs_fdr analysis. |
| Rmd | 9cd2a26 | Peter Carbonetto | 2022-01-25 | Small edit to power_vs_fdr analysis. |
| html | e5c993d | Peter Carbonetto | 2022-01-25 | Added in-sample power vs. fdr plot to power_vs_fdr analysis. |
| Rmd | ba7d922 | Peter Carbonetto | 2022-01-25 | workflowr::wflow_publish("analysis/power_vs_fdr.Rmd") |
| html | 0861415 | Peter Carbonetto | 2022-01-25 | Created power_vs_fdr page. |
| Rmd | 153840a | Peter Carbonetto | 2022-01-25 | workflowr::wflow_publish("analysis/power_vs_fdr.Rmd") |
| Rmd | b1ff774 | Peter Carbonetto | 2022-01-25 | Added new power_vs_fdr analysis file. |
Load some packages used to create the power vs. FDR plots:
library(ggplot2)
library(cowplot)Load the results from the UK Biobank simulations:
res0.005 = readRDS("data/roc_pve005.rds")
res0.1 = readRDS("data/roc_pve1.rds")
res0.3 = readRDS("data/roc_pve3.rds")Create the power vs. PDR plot comparing refinement procedure with the ‘in-sample’ LD:
pdat <-
rbind(data.frame(method = "without refinement",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineFALSE-ERTRUE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineFALSE-ERTRUE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineFALSE-ERTRUE-LDin"]]$pr[,3]),
data.frame(method = "with refinement",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,3]))
rows <- with(pdat,order(method,fdr))
pdat <- pdat[rows,]
p <- ggplot(pdat,aes(x = fdr,y = power,color = method,linetype = method)) +
geom_line(orientation = "y") +
geom_point(data = subset(pdat,thresh == 0.95),show.legend = FALSE) +
scale_color_manual(values = c("orange","blue")) +
scale_linetype_manual(values = c("solid","solid")) +
scale_x_continuous(breaks = seq(0,0.3,0.05)) +
scale_y_continuous(breaks = seq(0,0.3,0.05)) +
coord_cartesian(xlim = c(0,0.3),ylim = c(0,0.3)) +
theme_cowplot(font_size = 12) +
theme(panel.grid.major = element_line(colour = "gainsboro",size = 0.3))
print(p)
| Version | Author | Date |
|---|---|---|
| 02c5d23 | Yuxin Zou | 2022-06-20 |
Create the power vs. FDR plot comparing different fine-mapping methods when they are provided with the “in sample” LD:
pdat <-
rbind(data.frame(method = "SuSiE-RSS, est sigma",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,3]),
data.frame(method = "SuSiE-RSS, sigma=1",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDin"]]$pr[,3]),
data.frame(method = "SuSiE-RSS, sigma=1, L=true",
fdr = 1-res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDin"]]$pr[,3]),
data.frame(method = "FINEMAP",
fdr = 1-res0.005[["FINEMAP-L5-LDin"]]$pr[,1],
power = res0.005[["FINEMAP-L5-LDin"]]$pr[,2],
thresh = res0.005[["FINEMAP-L5-LDin"]]$pr[,3]),
data.frame(method = "FINEMAP, L=true",
fdr = 1-res0.005[["FINEMAP-Loracle-LDin"]]$pr[,1],
power = res0.005[["FINEMAP-Loracle-LDin"]]$pr[,2],
thresh = res0.005[["FINEMAP-Loracle-LDin"]]$pr[,3]),
data.frame(method = "DAP-G",
fdr = 1-res0.005[["DAP-G-LDin"]]$pr[,1],
power = res0.005[["DAP-G-LDin"]]$pr[,2],
thresh = res0.005[["DAP-G-LDin"]]$pr[,3]),
data.frame(method = "CAVIAR",
fdr = 1-res0.005[["CAVIAR-LDin"]]$pr[,1],
power = res0.005[["CAVIAR-LDin"]]$pr[,2],
thresh = res0.005[["CAVIAR-LDin"]]$pr[,3]))
rows <- with(pdat,order(method,fdr))
pdat <- pdat[rows,]
pdat$method = factor(pdat$method, levels = c('SuSiE-RSS, est sigma',
'SuSiE-RSS, sigma=1','SuSiE-RSS, sigma=1, L=true',
'FINEMAP', 'FINEMAP, L=true',
'DAP-G', 'CAVIAR'))
p <- ggplot(pdat,aes(x = fdr,y = power,color = method,linetype = method)) +
geom_line(orientation = "y") +
geom_point(data = subset(pdat,thresh == 0.95),show.legend = FALSE) +
scale_color_manual(values = c("black","magenta","magenta","darkorange",
"darkorange","dodgerblue","limegreen")) +
scale_linetype_manual(values = c("dotted","solid","dashed","solid",
"dashed","solid","solid")) +
scale_x_continuous(breaks = seq(0,0.3,0.05)) +
scale_y_continuous(breaks = seq(0,0.3,0.05)) +
coord_cartesian(xlim = c(0,0.3),ylim = c(0,0.3)) +
theme_cowplot(font_size = 12) +
theme(panel.grid.major = element_line(colour = "gainsboro",size = 0.3))
print(p)
Create the power vs. FDR plot comparing different fine-mapping methods when they are provided with the “out of sample” LD:
pdat1 <-
rbind(data.frame(method = "SuSiE-RSS (in-sample LD)",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,3]),
data.frame(method = "in-sample LD",
fdr = 1 - res0.005[["CAVIAR-LDin"]]$pr[,1],
power = res0.005[["CAVIAR-LDin"]]$pr[,2],
thresh = res0.005[["CAVIAR-LDin"]]$pr[,3]),
data.frame(method = "n=1000,lambda=0",
fdr = 1-res0.005[["CAVIAR-LDref1000"]]$pr[,1],
power = res0.005[["CAVIAR-LDref1000"]]$pr[,2],
thresh = res0.005[["CAVIAR-LDref1000"]]$pr[,3]),
data.frame(method = "n=1000,lambda=0.001",
fdr = 1 - res0.005[["CAVIAR-LDref1000-lambda0.001"]]$pr[,1],
power = res0.005[["CAVIAR-LDref1000-lambda0.001"]]$pr[,2],
thresh = res0.005[["CAVIAR-LDref1000-lambda0.001"]]$pr[,3]),
data.frame(method = "n=1000,lambda=estimated",
fdr = 1-res0.005[["CAVIAR-LDref1000-lambdaEstimate"]]$pr[,1],
power = res0.005[["CAVIAR-LDref1000-lambdaEstimate"]]$pr[,2],
thresh = res0.005[["CAVIAR-LDref1000-lambdaEstimate"]]$pr[,3]),
data.frame(method = "n=500,lambda=0",
fdr = 1 - res0.005[["CAVIAR-LDref500"]]$pr[,1],
power = res0.005[["CAVIAR-LDref500"]]$pr[,2],
thresh = res0.005[["CAVIAR-LDref500"]]$pr[,3]),
data.frame(method = "n=500,lambda=0.001",
fdr = 1 - res0.005[["CAVIAR-LDref500-lambda0.001"]]$pr[,1],
power = res0.005[["CAVIAR-LDref500-lambda0.001"]]$pr[,2],
thresh = res0.005[["CAVIAR-LDref500-lambda0.001"]]$pr[,3]),
data.frame(method = "n=500,lambda=estimated",
fdr = 1 - res0.005[["CAVIAR-LDref500-lambdaEstimate"]]$pr[,1],
power = res0.005[["CAVIAR-LDref500-lambdaEstimate"]]$pr[,2],
thresh = res0.005[["CAVIAR-LDref500-lambdaEstimate"]]$pr[,3]))
pdat2 <-
rbind(data.frame(method = "SuSiE-RSS (in-sample LD)",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,3]),
data.frame(method = "in-sample LD",
fdr = 1 - res0.005[["DAP-G-LDin"]]$pr[,1],
power = res0.005[["DAP-G-LDin"]]$pr[,2],
thresh = res0.005[["DAP-G-LDin"]]$pr[,3]),
data.frame(method = "n=1000,lambda=0",
fdr = 1-res0.005[["DAP-G-LDref1000"]]$pr[,1],
power = res0.005[["DAP-G-LDref1000"]]$pr[,2],
thresh = res0.005[["DAP-G-LDref1000"]]$pr[,3]),
data.frame(method = "n=1000,lambda=0.001",
fdr = 1 - res0.005[["DAP-G-LDref1000-lambda0.001"]]$pr[,1],
power = res0.005[["DAP-G-LDref1000-lambda0.001"]]$pr[,2],
thresh = res0.005[["DAP-G-LDref1000-lambda0.001"]]$pr[,3]),
data.frame(method = "n=1000,lambda=estimated",
fdr = 1-res0.005[["DAP-G-LDref1000-lambdaEstimate"]]$pr[,1],
power = res0.005[["DAP-G-LDref1000-lambdaEstimate"]]$pr[,2],
thresh = res0.005[["DAP-G-LDref1000-lambdaEstimate"]]$pr[,3]),
data.frame(method = "n=500,lambda=0",
fdr = 1 - res0.005[["DAP-G-LDref500"]]$pr[,1],
power = res0.005[["DAP-G-LDref500"]]$pr[,2],
thresh = res0.005[["DAP-G-LDref500"]]$pr[,3]),
data.frame(method = "n=500,lambda=0.001",
fdr = 1 - res0.005[["DAP-G-LDref500-lambda0.001"]]$pr[,1],
power = res0.005[["DAP-G-LDref500-lambda0.001"]]$pr[,2],
thresh = res0.005[["DAP-G-LDref500-lambda0.001"]]$pr[,3]),
data.frame(method = "n=500,lambda=estimated",
fdr = 1 - res0.005[["DAP-G-LDref500-lambdaEstimate"]]$pr[,1],
power = res0.005[["DAP-G-LDref500-lambdaEstimate"]]$pr[,2],
thresh = res0.005[["DAP-G-LDref500-lambdaEstimate"]]$pr[,3]))
pdat3 <-
rbind(data.frame(method = "SuSiE-RSS (in-sample LD)",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,3]),
data.frame(method = "in-sample LD",
fdr = 1 - res0.005[["FINEMAP-Loracle-LDin"]]$pr[,1],
power = res0.005[["FINEMAP-Loracle-LDin"]]$pr[,2],
thresh = res0.005[["FINEMAP-Loracle-LDin"]]$pr[,3]),
data.frame(method = "n=1000,lambda=0",
fdr = 1-res0.005[["FINEMAP-Loracle-LDref1000"]]$pr[,1],
power = res0.005[["FINEMAP-Loracle-LDref1000"]]$pr[,2],
thresh = res0.005[["FINEMAP-Loracle-LDref1000"]]$pr[,3]),
data.frame(method = "n=1000,lambda=0.001",
fdr = 1 - res0.005[["FINEMAP-Loracle-LDref1000-lambda0.001"]]$pr[,1],
power = res0.005[["FINEMAP-Loracle-LDref1000-lambda0.001"]]$pr[,2],
thresh = res0.005[["FINEMAP-Loracle-LDref1000-lambda0.001"]]$pr[,3]),
data.frame(method = "n=1000,lambda=estimated",
fdr = 1-res0.005[["FINEMAP-Loracle-LDref1000-lambdaEstimate"]]$pr[,1],
power = res0.005[["FINEMAP-Loracle-LDref1000-lambdaEstimate"]]$pr[,2],
thresh = res0.005[["FINEMAP-Loracle-LDref1000-lambdaEstimate"]]$pr[,3]),
data.frame(method = "n=500,lambda=0",
fdr = 1 - res0.005[["FINEMAP-Loracle-LDref500"]]$pr[,1],
power = res0.005[["FINEMAP-Loracle-LDref500"]]$pr[,2],
thresh = res0.005[["FINEMAP-Loracle-LDref500"]]$pr[,3]),
data.frame(method = "n=500,lambda=0.001",
fdr = 1 - res0.005[["FINEMAP-Loracle-LDref500-lambda0.001"]]$pr[,1],
power = res0.005[["FINEMAP-Loracle-LDref500-lambda0.001"]]$pr[,2],
thresh = res0.005[["FINEMAP-Loracle-LDref500-lambda0.001"]]$pr[,3]),
data.frame(method = "n=500,lambda=estimated",
fdr = 1 - res0.005[["FINEMAP-Loracle-LDref500-lambdaEstimate"]]$pr[,1],
power = res0.005[["FINEMAP-Loracle-LDref500-lambdaEstimate"]]$pr[,2],
thresh = res0.005[["FINEMAP-Loracle-LDref500-lambdaEstimate"]]$pr[,3]))
pdat4 <-
rbind(data.frame(method = "SuSiE-RSS (in-sample LD)",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,3]),
data.frame(method = "in-sample LD",
fdr = 1 - res0.005[["FINEMAP-L5-LDin"]]$pr[,1],
power = res0.005[["FINEMAP-L5-LDin"]]$pr[,2],
thresh = res0.005[["FINEMAP-L5-LDin"]]$pr[,3]),
data.frame(method = "n=1000,lambda=0",
fdr = 1-res0.005[["FINEMAP-L5-LDref1000"]]$pr[,1],
power = res0.005[["FINEMAP-L5-LDref1000"]]$pr[,2],
thresh = res0.005[["FINEMAP-L5-LDref1000"]]$pr[,3]),
data.frame(method = "n=1000,lambda=0.001",
fdr = 1 - res0.005[["FINEMAP-L5-LDref1000-lambda0.001"]]$pr[,1],
power = res0.005[["FINEMAP-L5-LDref1000-lambda0.001"]]$pr[,2],
thresh = res0.005[["FINEMAP-L5-LDref1000-lambda0.001"]]$pr[,3]),
data.frame(method = "n=1000,lambda=estimated",
fdr = 1-res0.005[["FINEMAP-L5-LDref1000-lambdaEstimate"]]$pr[,1],
power = res0.005[["FINEMAP-L5-LDref1000-lambdaEstimate"]]$pr[,2],
thresh = res0.005[["FINEMAP-L5-LDref1000-lambdaEstimate"]]$pr[,3]),
data.frame(method = "n=500,lambda=0",
fdr = 1 - res0.005[["FINEMAP-L5-LDref500"]]$pr[,1],
power = res0.005[["FINEMAP-L5-LDref500"]]$pr[,2],
thresh = res0.005[["FINEMAP-L5-LDref500"]]$pr[,3]),
data.frame(method = "n=500,lambda=0.001",
fdr = 1 - res0.005[["FINEMAP-L5-LDref500-lambda0.001"]]$pr[,1],
power = res0.005[["FINEMAP-L5-LDref500-lambda0.001"]]$pr[,2],
thresh = res0.005[["FINEMAP-L5-LDref500-lambda0.001"]]$pr[,3]),
data.frame(method = "n=500,lambda=estimated",
fdr = 1 - res0.005[["FINEMAP-L5-LDref500-lambdaEstimate"]]$pr[,1],
power = res0.005[["FINEMAP-L5-LDref500-lambdaEstimate"]]$pr[,2],
thresh = res0.005[["FINEMAP-L5-LDref500-lambdaEstimate"]]$pr[,3]))
pdat5 <-
rbind(data.frame(method = "SuSiE-RSS (in-sample LD)",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,3]),
data.frame(method = "in-sample LD",
fdr = 1 - res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDin"]]$pr[,3]),
data.frame(method = "n=1000,lambda=0",
fdr = 1-res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000"]]$pr[,1],
power = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000"]]$pr[,3]),
data.frame(method = "n=1000,lambda=0.001",
fdr = 1 - res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000-lambda0.001"]]$pr[,1],
power = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000-lambda0.001"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000-lambda0.001"]]$pr[,3]),
data.frame(method = "n=1000,lambda=estimated",
fdr = 1-res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000-lambdaEstimate"]]$pr[,1],
power = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000-lambdaEstimate"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000-lambdaEstimate"]]$pr[,3]),
data.frame(method = "n=500,lambda=0",
fdr = 1 - res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref500"]]$pr[,1],
power = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref500"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref500"]]$pr[,3]),
data.frame(method = "n=500,lambda=0.001",
fdr = 1 - res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref500-lambda0.001"]]$pr[,1],
power = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref500-lambda0.001"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref500-lambda0.001"]]$pr[,3]),
data.frame(method = "n=500,lambda=estimated",
fdr = 1 - res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref500-lambdaEstimate"]]$pr[,1],
power = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref500-lambdaEstimate"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref500-lambdaEstimate"]]$pr[,3]))
pdat6 <-
rbind(data.frame(method = "SuSiE-RSS (in-sample LD)",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,3]),
data.frame(method = "in-sample LD",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDin"]]$pr[,3]),
data.frame(method = "n=1000,lambda=0",
fdr = 1-res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000"]]$pr[,3]),
data.frame(method = "n=1000,lambda=0.001",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000-lambda0.001"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000-lambda0.001"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000-lambda0.001"]]$pr[,3]),
data.frame(method = "n=1000,lambda=estimated",
fdr = 1-res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000-lambdaEstimate"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000-lambdaEstimate"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000-lambdaEstimate"]]$pr[,3]),
data.frame(method = "n=500,lambda=0",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref500"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref500"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref500"]]$pr[,3]),
data.frame(method = "n=500,lambda=0.001",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref500-lambda0.001"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref500-lambda0.001"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref500-lambda0.001"]]$pr[,3]),
data.frame(method = "n=500,lambda=estimated",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref500-lambdaEstimate"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref500-lambdaEstimate"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref500-lambdaEstimate"]]$pr[,3]))
pdat <- rbind(cbind(row = 1,col = 1,pdat1),
cbind(row = 1,col = 2,pdat2),
cbind(row = 2,col = 1,pdat3),
cbind(row = 2,col = 2,pdat4),
cbind(row = 3,col = 1,pdat5),
cbind(row = 3,col = 2,pdat6))
rows <- with(pdat,order(row,col,method,fdr))
pdat <- pdat[rows,]
pdat$method = factor(pdat$method, levels = c('SuSiE-RSS (in-sample LD)',
'in-sample LD',
'n=1000,lambda=0',
'n=1000,lambda=0.001',
'n=1000,lambda=estimated',
'n=500,lambda=0',
'n=500,lambda=0.001',
'n=500,lambda=estimated'))
p <- ggplot(pdat,aes(x = fdr,y = power,color = method,linetype = method,
size = method)) +
facet_grid(rows = vars(row),cols = vars(col)) +
geom_line(orientation = "y") +
geom_point(data = subset(pdat,thresh == 0.95),
mapping = aes(x = fdr,y = power,color = method),
inherit.aes = FALSE,show.legend = FALSE) +
scale_color_manual(values = c("black","darkorange","royalblue","royalblue",
"royalblue","limegreen","limegreen",
"limegreen")) +
scale_linetype_manual(values = c("dotted","solid","solid","dotted","solid",
"solid","dotted","solid")) +
scale_size_manual(values = c(0.75,0.75,0.3,0.75,0.75,0.3,0.75,0.75)) +
scale_x_continuous(breaks = seq(0,0.3,0.05)) +
scale_y_continuous(breaks = seq(0,0.3,0.05)) +
coord_cartesian(xlim = c(0,0.3),ylim = c(0,0.3)) +
theme_cowplot(font_size = 10) +
theme(panel.grid.major = element_line(colour = "gainsboro",size = 0.3))
print(p)
Create the power vs. FDR plot comparing different fine-mapping methods when they are provided with the “out of sample” LD with higher PVE:
pdat1 <-
rbind(data.frame(method = "SuSiE-RSS, est sigma (in-sample LD)",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,3]),
data.frame(method = "SuSiE-RSS, sigma=1",
fdr = 1 - res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000"]]$pr[,1],
power = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000"]]$pr[,3]),
data.frame(method = "SuSiE-RSS, sigma=1, L=TRUE",
fdr = 1-res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000"]]$pr[,1],
power = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000"]]$pr[,2],
thresh = res0.005[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000"]]$pr[,3]),
data.frame(method = "FINEMAP",
fdr = 1 - res0.005[["FINEMAP-L5-LDref1000"]]$pr[,1],
power = res0.005[["FINEMAP-L5-LDref1000"]]$pr[,2],
thresh = res0.005[["FINEMAP-L5-LDref1000"]]$pr[,3]),
data.frame(method = "FINEMAP, L=TRUE",
fdr = 1-res0.005[["FINEMAP-Loracle-LDref1000"]]$pr[,1],
power = res0.005[["FINEMAP-Loracle-LDref1000"]]$pr[,2],
thresh = res0.005[["FINEMAP-Loracle-LDref1000"]]$pr[,3]),
data.frame(method = "DAP-G",
fdr = 1 - res0.005[["DAP-G-LDref1000"]]$pr[,1],
power = res0.005[["DAP-G-LDref1000"]]$pr[,2],
thresh = res0.005[["DAP-G-LDref1000"]]$pr[,3]),
data.frame(method = "CAVIAR",
fdr = 1 - res0.005[["CAVIAR-LDref1000"]]$pr[,1],
power = res0.005[["CAVIAR-LDref1000"]]$pr[,2],
thresh = res0.005[["CAVIAR-LDref1000"]]$pr[,3]))
pdat2 <-
rbind(data.frame(method = "SuSiE-RSS, est sigma (in-sample LD)",
fdr = 1 - res0.1[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,1],
power = res0.1[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,2],
thresh = res0.1[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,3]),
data.frame(method = "SuSiE-RSS, sigma=1",
fdr = 1 - res0.1[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000"]]$pr[,1],
power = res0.1[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000"]]$pr[,2],
thresh = res0.1[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000"]]$pr[,3]),
data.frame(method = "SuSiE-RSS, sigma=1, L=TRUE",
fdr = 1-res0.1[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000"]]$pr[,1],
power = res0.1[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000"]]$pr[,2],
thresh = res0.1[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000"]]$pr[,3]),
data.frame(method = "FINEMAP",
fdr = 1 - res0.1[["FINEMAP-L5-LDref1000"]]$pr[,1],
power = res0.1[["FINEMAP-L5-LDref1000"]]$pr[,2],
thresh = res0.1[["FINEMAP-L5-LDref1000"]]$pr[,3]),
data.frame(method = "FINEMAP, L=TRUE",
fdr = 1-res0.1[["FINEMAP-Loracle-LDref1000"]]$pr[,1],
power = res0.1[["FINEMAP-Loracle-LDref1000"]]$pr[,2],
thresh = res0.1[["FINEMAP-Loracle-LDref1000"]]$pr[,3]),
data.frame(method = "DAP-G",
fdr = 1 - res0.1[["DAP-G-LDref1000"]]$pr[,1],
power = res0.1[["DAP-G-LDref1000"]]$pr[,2],
thresh = res0.1[["DAP-G-LDref1000"]]$pr[,3]),
data.frame(method = "CAVIAR",
fdr = 1 - res0.1[["CAVIAR-LDref1000"]]$pr[,1],
power = res0.1[["CAVIAR-LDref1000"]]$pr[,2],
thresh = res0.1[["CAVIAR-LDref1000"]]$pr[,3]))
pdat3 <-
rbind(data.frame(method = "SuSiE-RSS, est sigma (in-sample LD)",
fdr = 1 - res0.3[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,1],
power = res0.3[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,2],
thresh = res0.3[["SuSiE-RSS-L10-refineTRUE-ERTRUE-LDin"]]$pr[,3]),
data.frame(method = "SuSiE-RSS, sigma=1",
fdr = 1 - res0.3[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000"]]$pr[,1],
power = res0.3[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000"]]$pr[,2],
thresh = res0.3[["SuSiE-RSS-L10-refineTRUE-ERFALSE-LDref1000"]]$pr[,3]),
data.frame(method = "SuSiE-RSS, sigma=1, L=TRUE",
fdr = 1-res0.3[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000"]]$pr[,1],
power = res0.3[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000"]]$pr[,2],
thresh = res0.3[["SuSiE-RSS-Ltrue-refineTRUE-ERFALSE-LDref1000"]]$pr[,3]),
data.frame(method = "FINEMAP",
fdr = 1 - res0.3[["FINEMAP-L5-LDref1000"]]$pr[,1],
power = res0.3[["FINEMAP-L5-LDref1000"]]$pr[,2],
thresh = res0.3[["FINEMAP-L5-LDref1000"]]$pr[,3]),
data.frame(method = "FINEMAP, L=TRUE",
fdr = 1-res0.3[["FINEMAP-Loracle-LDref1000"]]$pr[,1],
power = res0.3[["FINEMAP-Loracle-LDref1000"]]$pr[,2],
thresh = res0.3[["FINEMAP-Loracle-LDref1000"]]$pr[,3]),
data.frame(method = "DAP-G",
fdr = 1 - res0.3[["DAP-G-LDref1000"]]$pr[,1],
power = res0.3[["DAP-G-LDref1000"]]$pr[,2],
thresh = res0.3[["DAP-G-LDref1000"]]$pr[,3]),
data.frame(method = "CAVIAR",
fdr = 1 - res0.3[["CAVIAR-LDref1000"]]$pr[,1],
power = res0.3[["CAVIAR-LDref1000"]]$pr[,2],
thresh = res0.3[["CAVIAR-LDref1000"]]$pr[,3]))
pdat <- rbind(cbind(row = 1,col = 0.005,pdat1),
cbind(row = 1,col = 0.1,pdat2),
cbind(row = 1,col = 0.3,pdat3))
rows <- with(pdat,order(row,col,method,fdr))
pdat <- pdat[rows,]
pdat$method = factor(pdat$method, levels = c('SuSiE-RSS, est sigma (in-sample LD)',
'SuSiE-RSS, sigma=1',
'SuSiE-RSS, sigma=1, L=TRUE',
'FINEMAP',
'FINEMAP, L=TRUE',
'DAP-G',
'CAVIAR'))
p <- ggplot(pdat,aes(x = fdr,y = power,color = method,linetype = method)) +
facet_grid(rows = vars(row),cols = vars(col)) +
geom_line(orientation = "y") +
geom_point(data = subset(pdat,thresh == 0.95),
mapping = aes(x = fdr,y = power,color = method),
inherit.aes = FALSE,show.legend = FALSE) +
scale_color_manual(values = c("black","magenta","magenta","darkorange","darkorange",
"royalblue","limegreen")) +
scale_linetype_manual(values = c("dotted","solid",'dashed',"solid","dashed",
"solid","solid")) +
scale_x_continuous(breaks = seq(0,0.3,0.05)) +
scale_y_continuous(breaks = seq(0,0.3,0.05)) +
coord_cartesian(xlim = c(0,0.3),ylim = c(0,0.3)) +
theme_cowplot(font_size = 10) +
theme(panel.grid.major = element_line(colour = "gainsboro",size = 0.3))
print(p)
| Version | Author | Date |
|---|---|---|
| 02c5d23 | Yuxin Zou | 2022-06-20 |
sessionInfo()
# R version 4.1.0 (2021-05-18)
# Platform: x86_64-apple-darwin17.0 (64-bit)
# Running under: macOS Big Sur 10.16
#
# Matrix products: default
# BLAS: /Library/Frameworks/R.framework/Versions/4.1/Resources/lib/libRblas.dylib
# LAPACK: /Library/Frameworks/R.framework/Versions/4.1/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] cowplot_1.1.1 ggplot2_3.3.6 workflowr_1.7.0
#
# loaded via a namespace (and not attached):
# [1] tidyselect_1.1.2 xfun_0.30 bslib_0.3.1 purrr_0.3.4
# [5] colorspace_2.0-3 vctrs_0.4.1 generics_0.1.2 htmltools_0.5.2
# [9] yaml_2.3.5 utf8_1.2.2 rlang_1.0.2 jquerylib_0.1.4
# [13] later_1.3.0 pillar_1.7.0 glue_1.6.2 withr_2.5.0
# [17] DBI_1.1.2 lifecycle_1.0.1 stringr_1.4.0 munsell_0.5.0
# [21] gtable_0.3.0 ragg_1.2.2 evaluate_0.15 knitr_1.39
# [25] callr_3.7.0 fastmap_1.1.0 httpuv_1.6.5 ps_1.7.0
# [29] fansi_1.0.3 highr_0.9 Rcpp_1.0.8.3 promises_1.2.0.1
# [33] scales_1.2.0 jsonlite_1.8.0 systemfonts_1.0.4 farver_2.1.0
# [37] fs_1.5.2 textshaping_0.3.6 digest_0.6.29 stringi_1.7.6
# [41] processx_3.5.3 dplyr_1.0.9 getPass_0.2-2 rprojroot_2.0.3
# [45] grid_4.1.0 cli_3.3.0 tools_4.1.0 magrittr_2.0.3
# [49] sass_0.4.1 tibble_3.1.7 crayon_1.5.1 whisker_0.4
# [53] pkgconfig_2.0.3 ellipsis_0.3.2 assertthat_0.2.1 rmarkdown_2.14
# [57] httr_1.4.3 rstudioapi_0.13 R6_2.5.1 git2r_0.30.1
# [61] compiler_4.1.0