start.pos="random.f500"

snpgdsLDpruning()

DESCRIPTION

@@ -3,7 +3,7 @@ Type: Package
 Title: Parallel Computing Toolset for Relatedness and Principal Component
         Analysis of SNP Data
 Version: 1.37.2
-Date: 2024-03-08
+Date: 2024-03-26
 Depends: R (>= 2.15), gdsfmt (>= 1.8.3)
 LinkingTo: gdsfmt
 Imports: methods

NEWS

@@ -7,6 +7,10 @@ UTILITIES
 
     o new option 'autosave' in `snpgdsLDpruning()`
 
+    o new option `start.pos="random.f500"` in `snpgdsLDpruning()`, allowing
+      faster genotype scanning (it is used by default now). To be compatible
+      with the previous versions, please specify `start.pos="random"` manually
+
     o Fix the welcome message when running in a non x86 system
 
 

R/LD.R

@@ -100,7 +100,8 @@ snpgdsLDMat <- function(gdsobj, sample.id=NULL, snp.id=NULL, slide=250L,
 snpgdsLDpruning <- function(gdsobj, sample.id=NULL, snp.id=NULL,
     autosome.only=TRUE, remove.monosnp=TRUE, maf=0.005, missing.rate=0.05,
     method=c("composite", "r", "dprime", "corr"), slide.max.bp=500000L,
-    slide.max.n=NA, ld.threshold=0.2, start.pos=c("random", "first", "last"),
+    slide.max.n=NA, ld.threshold=0.2,
+    start.pos=c("random.f500", "random", "first", "last"),
     num.thread=1L, autosave=NULL, verbose=TRUE)
 {
     # check
@@ -123,9 +124,8 @@ snpgdsLDpruning <- function(gdsobj, sample.id=NULL, snp.id=NULL,
         if (length(autosave)!=1L || anyNA(autosave) || autosave=="")
             stop("'autosave' should be NULL or a file name.")
     }
-    stopifnot(is.logical(verbose), length(verbose)==1L)
-
     start.pos <- match.arg(start.pos)
+    stopifnot(is.logical(verbose), length(verbose)==1L)
 
     if (verbose)
     {
@@ -196,6 +196,7 @@ snpgdsLDpruning <- function(gdsobj, sample.id=NULL, snp.id=NULL,
 
             # call LD prune for this chromosome
             startidx <- switch(start.pos,
+                random.f500 = sample.int(min(n.tmp, 500L), 1L),
                 random = sample.int(n.tmp, 1L),
                 first  = 1L,
                 last   = n.tmp)

man/snpgdsLDpruning.Rd

@@ -10,7 +10,8 @@
 snpgdsLDpruning(gdsobj, sample.id=NULL, snp.id=NULL, autosome.only=TRUE,
     remove.monosnp=TRUE, maf=0.005, missing.rate=0.05,
     method=c("composite", "r", "dprime", "corr"), slide.max.bp=500000L,
-    slide.max.n=NA, ld.threshold=0.2, start.pos=c("random", "first", "last"),
+    slide.max.n=NA, ld.threshold=0.2,
+    start.pos=c("random.f500", "random", "first", "last"),
     num.thread=1L, autosave=NULL, verbose=TRUE)
 }
 \arguments{
@@ -31,8 +32,10 @@ snpgdsLDpruning(gdsobj, sample.id=NULL, snp.id=NULL, autosome.only=TRUE,
     \item{slide.max.bp}{the maximum basepairs in the sliding window}
     \item{slide.max.n}{the maximum number of SNPs in the sliding window}
     \item{ld.threshold}{the LD threshold}
-    \item{start.pos}{"random": a random starting position; "first": start from
-        the first position; "last": start from the last position}
+    \item{start.pos}{"random.f500", a starting postion randomly selected from
+        the first 500 markers (by default); "random": a random starting
+        position; "first": start from the first position; "last": start from
+        the last position. "random.f500" is applicable for >= v1.37.2}
     \item{num.thread}{the number of (CPU) cores used; if \code{NA}, detect
         the number of cores automatically}
     \item{autosave}{NULL or a file name for autosaving a single R object