[Gtdb-commits] r55 - in pkg/gt.db: R src

[noreply at r-forge.r-project.org]

Author: dahinds
Date: 2010-06-01 23:18:50 +0200 (Tue, 01 Jun 2010)
New Revision: 55

Modified:
   pkg/gt.db/R/hapmap.R
   pkg/gt.db/src/encode.c
Log:
- update to handle deletion/insertion polymorphisms in hapmap files



Modified: pkg/gt.db/R/hapmap.R
===================================================================
--- pkg/gt.db/R/hapmap.R	2010-06-01 20:33:59 UTC (rev 54)
+++ pkg/gt.db/R/hapmap.R	2010-06-01 21:18:50 UTC (rev 55)
@@ -24,9 +24,10 @@
 
 .fixup.hapmap.alleles <- function(alleles, geno)
 {
-    # fill in non-SNP allele lists based on the actual data
-    bad <- grep('^[ACGT]/[ACGT]$', alleles, invert=TRUE)
-    if (!length(bad)) return(alleles)
+    # fill in non-bi-allelic allele lists based on the actual data
+    bad <- (!grepl('^[ACGT]/[ACGT]$', alleles) &
+            !grepl('^-/[ACGT]+$', alleles))
+    if (!any(bad)) return(alleles)
     gt <- c('A','C','G','T')
     tbl <- (ch.table(geno[bad], chars=gt) > 0)
     a12 <- apply(tbl, 1, function(x) c(gt[x],'?','?')[1:2])

Modified: pkg/gt.db/src/encode.c
===================================================================
--- pkg/gt.db/src/encode.c	2010-06-01 20:33:59 UTC (rev 54)
+++ pkg/gt.db/src/encode.c	2010-06-01 21:18:50 UTC (rev 55)
@@ -61,6 +61,11 @@
 	for (i = 0; i < LENGTH(sa); i++) {
 		a = CHAR(STRING_ELT(sa,i));
 		a1 = a[0]; a2 = a[2];
+		if (a1 == '-') {
+			/* codes for insertion/deletion polymorphisms */
+			a1 = 'D';
+			a2 = 'I';
+		}
 		g = CHAR(STRING_ELT(sg,i));
 		for (j = 0; j < len && *g; j++) {
 			buf[j] = '?';