snpgdsLDpruning: show a progress bar

NEWS

@@ -5,6 +5,11 @@ UTILITIES
 
     o fix the compiler warning: -Wformat-security
 
+    o set 'maf=0.005' & 'missing.rate=0.05' by default in `snpgdsLDpruning()`
+      (for WGS data)
+
+    o show a progress bar in `snpgdsLDpruning()` when 'verbose=TRUE'
+
 
 CHANGES IN VERSION 1.34.0
 -------------------------

R/LD.R

@@ -98,10 +98,10 @@ 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=NaN, missing.rate=NaN,
-    method=c("composite", "r", "dprime", "corr"),
-    slide.max.bp=500000L, slide.max.n=NA, ld.threshold=0.2,
-    start.pos=c("random", "first", "last"), num.thread=1L, verbose=TRUE)
+    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"),
+    num.thread=1L, verbose=TRUE)
 {
     # check
     ws <- .InitFile2(
@@ -199,7 +199,7 @@ snpgdsLDpruning <- function(gdsobj, sample.id=NULL, snp.id=NULL,
                 first  = 1L,
                 last   = n.tmp)
             rv <- .Call(gnrLDpruning, startidx, position[flag],
-                slide.max.bp, slide.max.n, ld.threshold, method)
+                slide.max.bp, slide.max.n, ld.threshold, method, verbose)
 
             # output
             L <- rep(FALSE, length(total.snp.ids))

man/snpgdsLDpruning.Rd

@@ -8,7 +8,7 @@
 }
 \usage{
 snpgdsLDpruning(gdsobj, sample.id=NULL, snp.id=NULL, autosome.only=TRUE,
-    remove.monosnp=TRUE, maf=NaN, missing.rate=NaN,
+    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"),
     num.thread=1L, verbose=TRUE)

src/dGenGWAS.cpp

@@ -1544,7 +1544,6 @@ CdProgression::CdProgression(int type, bool show)
 {
 	fShow = show;
 	Init(0, false);
-
 	switch (fType = type)
 	{
 	case 0:
@@ -1553,6 +1552,9 @@ CdProgression::CdProgression(int type, bool show)
 	case 1:
 		TimeInterval = 5 * CLOCKS_PER_SEC;
 		break;
+	case 2:
+		TimeInterval = 15 * CLOCKS_PER_SEC;
+		break;
 	}
 }
 
@@ -1571,6 +1573,7 @@ void CdProgression::Init(C_Int64 TotalCnt, bool ShowInit)
 	fTotal = TotalCnt;
 	fCurrent = fPercent = 0;
 	OldTime = clock();
+	current_bar = 0;
 	if (ShowInit) ShowProgress();
 }
 
@@ -1618,6 +1621,16 @@ void CdProgression::ShowProgress()
 				// fflush(stdout);
 				break;
 			}
+		case 2:
+			{
+				int n = (int)round(fPercent * 0.20);
+				if (n > current_bar)
+				{
+					string s(n - current_bar, '=');
+					Rprintf("%s", s.c_str());
+					current_bar = n;
+				}
+			}
 		}
 	}
 }

src/dGenGWAS.h

@@ -2,7 +2,7 @@
 //
 // dGenGWAS.h: Workspace of Genome-Wide Association Studies
 //
-// Copyright (C) 2011-2017    Xiuwen Zheng
+// Copyright (C) 2011-2024    Xiuwen Zheng
 //
 // This file is part of SNPRelate.
 //
@@ -414,6 +414,8 @@ namespace GWAS
 		int fPercent;
 		bool fShow;
 		clock_t TimeInterval, OldTime;
+	private:
+		int current_bar;  // used when type=2
 	};
 
 

src/genLD.cpp

@@ -2,7 +2,7 @@
 //
 // genLD.cpp: Linkage Disequilibrium (LD) Analysis on GWAS
 //
-// Copyright (C) 2011-2018    Xiuwen Zheng
+// Copyright (C) 2011-2024    Xiuwen Zheng
 //
 // This file is part of SNPRelate.
 //
@@ -753,7 +753,7 @@ namespace LD
 
 	COREARRAY_DLL_LOCAL void Perform_LD_Pruning(int StartIdx, int *pos_bp,
 		int slide_max_bp, int slide_max_n, const double LD_threshold,
-		C_BOOL *out_SNP)
+		C_BOOL *out_SNP, bool verbose)
 	{
 		// initial variables
 		nPackedSamp = (MCWorkingGeno.Space().SampleNum() % 4 > 0) ?
@@ -769,12 +769,16 @@ namespace LD
 		CdBufSpace BufSNP(MCWorkingGeno.Space(), true, CdBufSpace::acInc);
 		ListGeno.push_back(TSNP(StartIdx, pos_bp[StartIdx], nPackedSamp));
 		BufSNP.ReadPackedGeno(StartIdx, &(ListGeno.back().genobuf[0]));
+		CdProgression progress(2, verbose);
+		if (verbose) Rprintf("|");
 
 		// for-loop, increasing
+		progress.Init(MCWorkingGeno.Space().SNPNum()-StartIdx-1, verbose);
 		for (int i=StartIdx+1; i < MCWorkingGeno.Space().SNPNum(); i++)
 		{
 			// load genotypes
 			BufSNP.ReadPackedGeno(i, &buf[0]);
+			progress.Forward(1, verbose);
 			// detect LD
 			int TotalCnt = 0, ValidCnt = 0;
 			list<TSNP>::iterator it;
@@ -807,12 +811,13 @@ namespace LD
 		// -----------------------------------------------------
 		// decreasing searching: i --> i-1
 
+		if (verbose) Rprintf("|");
 		ListGeno.clear();
 		for (int i=StartIdx; i < MCWorkingGeno.Space().SNPNum(); i++)
 		{
 			if (out_SNP[i])
 			{
-				// check whether it is in the sliding window
+				// check whether it is in the sliding window of StartIdx
 				if ((abs(i - StartIdx) <= slide_max_n) &&
 					(abs(pos_bp[i] - pos_bp[StartIdx]) <= slide_max_bp))
 				{
@@ -824,11 +829,13 @@ namespace LD
 		}
 
 		BufSNP.SetAccessFlag(CdBufSpace::acDec);
-		// for-loop, descreasing
+		progress.Init(StartIdx-1, verbose);
+		// for-loop, decreasing
 		for (int i=StartIdx-1; i >= 0; i--)
 		{
 			// load genotypes
 			BufSNP.ReadPackedGeno(i, &buf[0]);
+			progress.Forward(1, verbose);
 			// detect LD
 			int TotalCnt = 0, ValidCnt = 0;
 			list<TSNP>::iterator it;
@@ -857,6 +864,8 @@ namespace LD
 				memmove(&(ListGeno.front().genobuf[0]), &buf[0], nPackedSamp);
 			}
 		}
+
+		if (verbose) Rprintf("|\n");
 	}
 }
 
@@ -947,15 +956,17 @@ COREARRAY_DLL_EXPORT SEXP gnrLDMat(SEXP method, SEXP NumSlide, SEXP MatTrim,
 
 /// Prune SNPs based on LD
 COREARRAY_DLL_EXPORT SEXP gnrLDpruning(SEXP StartIdx, SEXP pos_bp,
-	SEXP slide_max_bp, SEXP slide_max_n, SEXP LD_threshold, SEXP method)
+	SEXP slide_max_bp, SEXP slide_max_n, SEXP LD_threshold, SEXP method,
+	SEXP verbose)
 {
 	COREARRAY_TRY
 
 		vector<C_BOOL> flag(MCWorkingGeno.Space().SNPNum());
 		LD::LD_Method = Rf_asInteger(method);
 		LD::Perform_LD_Pruning(Rf_asInteger(StartIdx)-1, INTEGER(pos_bp),
 			Rf_asInteger(slide_max_bp), Rf_asInteger(slide_max_n),
-			Rf_asReal(LD_threshold), &flag[0]);
+			Rf_asReal(LD_threshold), &flag[0],
+			Rf_asLogical(verbose) == TRUE);
 
 		PROTECT(rv_ans = NEW_LOGICAL(MCWorkingGeno.Space().SNPNum()));
 		int *p = LOGICAL(rv_ans);