add vignettes

.Rbuildignore

@@ -15,3 +15,4 @@
 ^README_cache$
 ^README_files$
 ^LICENSE\.md$
+^data-raw$

DESCRIPTION

@@ -1,7 +1,7 @@
 Type: Package
 Package: CASE
 Title: Cell-type-specific And Shared EQTL fine-mapping
-Version: 0.2.1
+Version: 0.2.2
 Authors@R: c(
     person("Chen", "Lin", , "[email protected]", role = c("aut", "cre"),
            comment = c(ORCID = "0000-0001-9821-2578")),
@@ -16,12 +16,18 @@ Imports:
     magrittr,
     MASS,
     mvtnorm,
-    stats,
-    CorBin,
-    mvsusieR,
-    susieR
+    stats
 Encoding: UTF-8
 RoxygenNote: 7.3.1
 Suggests: 
-    testthat (>= 3.0.0)
+    knitr,
+    rmarkdown,
+    testthat (>= 3.0.0),
+    scales,
+    ggplot2,
+    susieR
 Config/testthat/edition: 3
+Depends: 
+    R (>= 2.10)
+LazyData: true
+VignetteBuilder: knitr

NAMESPACE

@@ -4,21 +4,14 @@ export(CASE)
 export(CASE_test)
 export(CASE_train)
 export(get_credible_sets)
-importFrom(CorBin,cBern)
 importFrom(MASS,ginv)
 importFrom(magrittr,"%>%")
-importFrom(mvsusieR,create_mixture_prior)
-importFrom(mvsusieR,mvsusie_rss)
 importFrom(mvtnorm,rmvnorm)
-importFrom(stats,cor)
 importFrom(stats,cov2cor)
-importFrom(stats,lm)
 importFrom(stats,p.adjust)
 importFrom(stats,pnorm)
 importFrom(stats,qchisq)
 importFrom(stats,qnorm)
 importFrom(stats,quantile)
 importFrom(stats,rmultinom)
-importFrom(stats,rnorm)
 importFrom(stats,sd)
-importFrom(susieR,susie_rss)

R/data.R

@@ -0,0 +1,12 @@
+#' Example Data
+#'
+#' An example data of three traits for the CASE fine-mapping illustration.
+#'
+#' @format ## `example_data`
+#' A list contains 3 elements:
+#' \describe{
+#'   \item{Y}{500 by 3 matrix of phenotype.}
+#'   \item{X}{500 by 1000 matrix of genotype. This is created by R package `CorBin`}
+#'   \item{B}{1000 by 3 matrix of eQTL effects. Only the 10th and 950th SNPs have eQTL effects.}
+#' }
+"example_data"

README.md

@@ -1,14 +1,20 @@
 # CASE: Cell-type-specific And Shared EQTL Fine-mapping
 
-`CASE` is a R package for multi-trait fine-mapping analysis, focusing on single-cell eQTL fine-mapping.
+`CASE` is an R package designed for multi-trait fine-mapping analysis, with a particular focus on single-cell eQTL fine-mapping.
 
 ## 1. Installation
 
 ``` r
-#install.packages("devtools")
+# Install the devtools package if you haven't already
+# install.packages("devtools")
+
 devtools::install_github("leaffur/CASE")
 ```
 
 ## 2. Quick Start
 
-After `library(CASE)` in R, please use `?CASE` command for simple examples.
+For an introductory overview, please refer to the vignettes provided in the package.
+
+After loading the package with `library(CASE)`, you can access simple examples and guidance using the `?CASE` command in R.
+
+This is an early version of CASE, and we welcome any feedback or suggestions for improvement.

data-raw/example_data.R

@@ -0,0 +1,24 @@
+## code to prepare `example_data` dataset goes here
+library(CorBin)
+
+set.seed(1024)
+N = 500
+M = 1000
+C = 3
+
+X = cbind(cBern(N, rep(0.3, M/4), rho = 0.95, type = "DCP"),
+          cBern(N, rep(0.3, M/2), rho = 0.85, type = "DCP"),
+          cBern(N, rep(0.3, M/4), rho = 0.98, type = "DCP"))
+
+B = matrix(0, M, C)
+idx1 = 10
+idx2 = 950
+B[idx1, ] = c(sqrt(0.3), sqrt(0.2), sqrt(0.1))
+B[idx2, ] = c(sqrt(0.25), sqrt(0.15), 0)
+
+e = matrix(rnorm(N * C), N, C)
+Y = X %*% B + e
+
+example_data = list(Y = Y, X = X, B = B)
+
+usethis::use_data(example_data, overwrite = TRUE)

vignettes/.gitignore

@@ -0,0 +1,2 @@
+*.html
+*.R

vignettes/Introduction_to_CASE.Rmd

@@ -0,0 +1,104 @@
+---
+title: "Introduction to CASE for multi-trait fine-mapping"
+output: rmarkdown::html_vignette
+vignette: >
+  %\VignetteIndexEntry{Introduction_to_CASE}
+  %\VignetteEngine{knitr::rmarkdown}
+  %\VignetteEncoding{UTF-8}
+---
+
+```{r, include = FALSE}
+knitr::opts_chunk$set(
+  collapse = TRUE,
+  comment = "#>"
+)
+```
+
+This Vignette is a draft and will be further updated.
+
+First, load necessary packages.
+
+```{r setup}
+library(CASE)
+library(ggplot2)
+library(susieR)
+```
+
+```{r}
+data("example_data")
+X = example_data$X
+Y = example_data$Y
+B = example_data$B
+
+N = nrow(X)
+M = ncol(X)
+C = ncol(Y)
+cat("sample size =", N, "\n",
+    "SNP size =", M, "\n",
+    "Cell type number =", C, "\n")
+```
+
+Let's have a look at how the Linkage Disequilibrium (LD) of the example data looks like.
+
+```{r}
+R = cor(X)
+
+df <- expand.grid(seq(M), seq(M))
+df$corr <- as.vector(R)
+
+g2 = ggplot(df, aes(x = Var1, y = Var2, fill = corr)) +
+  geom_tile() +
+  scale_fill_gradient2(low = "blue", mid = "white", high = "red", limits = c(-1, 1),
+                       midpoint = 0, na.value = NA,
+                       guide = guide_colorbar(
+                         barwidth = 1, barheight = 7, title.position = "top", oob = scales::oob_squish
+                       )
+                       ) +
+  xlab("") + 
+  ylab("") +
+  theme(aspect.ratio = 1, axis.text.x = element_text(angle = 90, hjust = 1, vjust = 0.5),
+        strip.background = element_rect(fill = "white", color = "white"),
+        panel.border = element_rect(colour = "black", fill = NA, 
+                                    linewidth = 1),
+        strip.text = element_text(color = "black", size = 20),
+        text = element_text(size = 20)) +
+  labs(fill = "corr")
+print(g2)
+```
+
+```{r}
+print(which(B != 0, arr.ind = TRUE))
+idx1 = 10
+idx2 = 950
+B[c(idx1, idx2), ]
+```
+
+```{r}
+Z = matrix(0, M, C)
+for (i in 1:M){
+  m1 = summary(lm(Y ~ X[, i]))
+  Z[i, ] = sapply(m1, function(x) x$coefficients[2, 3])
+}
+Z[c(idx1, idx2), ] 
+```
+
+Run SuSiE (a single-trait fine-mapping method) seperately.
+
+```{r}
+for (c in 1:C){
+  f1 = susie_rss(z = Z[, c], R = R, n = N)
+  print(f1$sets)
+}
+```
+
+Jointly studying three traits together improves the power of identifying the causal variants.
+
+```{r}
+fit <- CASE(Z = Z, R = R, N = rep(N, C))
+print(fit$sets)
+```
+
+
+```{r}
+sessionInfo()
+```