More revisions to the roxygen2 docs.

DESCRIPTION

@@ -15,7 +15,7 @@ Description: Implements methods for variable selection in linear
     and fast, allowing the SuSiE model be fit to large data sets
     (thousands of samples and hundreds of thousands of variables).
 Date: 2020-06-30
-Version: 0.9.7
+Version: 0.9.8
 Authors@R: c(person("Gao","Wang",role="aut",email="[email protected]"),
              person("Yuxin","Zou",role="aut"),
              person("Kaiqian","Zhang",role="aut"),

NAMESPACE

@@ -27,9 +27,11 @@ export(susie_plot_iteration)
 export(susie_rss)
 export(susie_suff_stat)
 export(susie_trendfilter)
+importFrom(Matrix,colMeans)
 importFrom(Matrix,colSums)
 importFrom(Matrix,crossprod)
 importFrom(Matrix,diag)
+importFrom(Matrix,rowSums)
 importFrom(Matrix,t)
 importFrom(Matrix,tcrossprod)
 importFrom(grDevices,dev.off)

R/set_R_attributes.R

@@ -1,8 +1,9 @@
-#' @title sets the attributes for the R matrix
-#' @param R a p by p LD matrix
-#' @param r_tol tolerance level for eigen value check of positive semidefinite matrix of R
-#' @return R with attribute e.g.
-#'         attr(R, 'eigenR') is the eigen decomposition of R.
+# @title sets the attributes for the R matrix
+# @param R a p by p LD matrix
+# @param r_tol tolerance level for eigen value check of positive
+#   semidefinite matrix of R.
+# @return R with attribute e.g., attr(R, 'eigenR') is the eigen
+#   decomposition of R.
 set_R_attributes = function(R, r_tol){
   if(is.null(attr(R, 'eigen')))
     eigenR = eigen(R, symmetric = TRUE)

R/set_X_attributes.R

@@ -1,15 +1,19 @@
-#' @title sets three attributes for matrix X
-#' @param X an n by p data matrix that can be either a trend filtering matrix or a regular dense/sparse matrix
-#' @param center boolean indicating centered by column means or not
-#' @param scale boolean indicating scaled by column standard deviations or not
-#' @return X with three attributes e.g.
-#'         attr(X, 'scaled:center') is a p vector of column means of X if center=TRUE, a p vector of 0s otherwise.
-#'         attr(X, 'scaled:scale') is a p vector of column standard deviations of X if scale=TRUE, a p vector of 1s otherwise.
-#'         attr(X, 'd') is a p vector of column sums of X.standardized^2,
-#'         where X.standardized is the matrix X centered by attr(X, 'scaled:center') and scaled by attr(X, 'scaled:scale').
-set_X_attributes = function(X,
-                             center = TRUE,
-                             scale = TRUE) {
+# @title sets three attributes for matrix X
+# @param X an n by p data matrix that can be either a trend filtering
+#   matrix or a regular dense/sparse matrix
+# @param center boolean indicating centered by column means or not
+# @param scale boolean indicating scaled by column standard deviations or not
+# @return X with three attributes e.g. `attr(X, 'scaled:center') is a
+# p vector of column means of X if center=TRUE, a p vector of zeros
+# otherwise. 'attr(X, 'scaled:scale') is a p vector of columan standard
+# deviations of X if scale=TRUE, a p vector of 1s otherwise. 'attr(X,
+# 'd') is a p vector of column sums of X.standardized^2,' where
+# X.standardized is the matrix X centered by attr(X, 'scaled:center')
+# and scaled by attr(X, 'scaled:scale').
+#
+#' @importFrom Matrix rowSums
+#' @importFrom Matrix colMeans
+set_X_attributes = function(X, center = TRUE, scale = TRUE) {
   # if X is a trend filtering matrix
   if (!is.null(attr(X, "matrix.type"))) {
     order <- attr(X,"order")
@@ -18,40 +22,36 @@ set_X_attributes = function(X,
     attr(X, "scaled:center") <- compute_tf_cm(order, n)
     attr(X, "scaled:scale") <- compute_tf_csd(order, n)
     attr(X, "d") <- compute_tf_d(order,n,attr(X, "scaled:center"),attr(X, "scaled:scale"),scale,center)
-    if (!center) {
+    if (!center)
       attr(X, "scaled:center") <- rep(0, n)
-    }
-    if (!scale) {
+    if (!scale)
       attr(X, "scaled:scale") <- rep(1, n)
-    }
   } else {
     # if X is either a dense or sparse ordinary matrix
     # get column means
-    cm = Matrix::colMeans(X, na.rm = TRUE)
+    cm = colMeans(X, na.rm = TRUE)
     # get column standard deviations
     csd = compute_colSds(X)
     # set sd = 1 when the column has variance 0
     csd[csd == 0] = 1
-    if (!center) {
+    if (!center)
       cm = rep(0, length = length(cm))
-    }
-    if (!scale) {
+    if (!scale) 
       csd = rep(1, length = length(cm))
-    }
     X.std = (t(X) - cm) / csd
     # set three attributes for X
-    attr(X, "d") <- Matrix::rowSums(X.std * X.std)
+    attr(X, "d") <- rowSums(X.std * X.std)
     attr(X, "scaled:center") <- cm
     attr(X, "scaled:scale") <- csd
   }
   return(X)
 }
 
-
-#' @title computes column standard deviations for any type of matrix
-#'        replace matrixStats::colSds since this function only takes a dense matrix
-#' @param X an n by p matrix of any type, e.g. sparse, dense
-#' @return a p vector of column standard deviations
+# @title computes column standard deviations for any type of matrix
+# @details Replace matrixStats::colSds since this function only takes
+#   a dense matrix.
+# @param X an n by p matrix of any type, e.g. sparse, dense.
+# @return a p vector of column standard deviations.
 compute_colSds = function(X){
   n = nrow(X)
   return(sqrt((colSums(X^2)/n - (colSums(X)/n)^2)*(n/(n-1))))