Minor tweaks

R/helpers.R

@@ -1,15 +1,15 @@
 #' @keywords internal
 #' @importFrom Biostrings IUPAC_CODE_MAP
 #' @importFrom S4Vectors mcols mcols<-
-.checkAlts <- function(var, alt_col) {
+.checkAlts <- function(var, alt_col, ref_col = "REF") {
 
     alt_col <- match.arg(alt_col, colnames(mcols(var)))
     alts <- mcols(var)[[alt_col]]
     if (is(alts, "XStringSetList")) alts <- as.character(unlist(alts))
 
     ## Check duplicate loci
-    dup <- duplicated(var)
     err <- c()
+    dup <- duplicated(var)
     if (any(dup))
         err <- c(
             err, "Duplicate variant loci found. Please choose which one to use"

R/varTypes.R

@@ -37,9 +37,10 @@
 #' @importFrom IRanges width
 #' @export
 varTypes <- function(x, alt_col = "ALT", ...){
+
     ## x should be a GRanges object with variants
     stopifnot(is(x, "GRanges"))
-    stopifnot(alt_col %in% colnames(mcols(x)))
+    x <- .checkAlts(x, alt_col)
 
     ## Basic info
     w <- width(x)
@@ -53,11 +54,11 @@ varTypes <- function(x, alt_col = "ALT", ...){
 
     ## Classify each range
     snv <- w == 1 & alt_width == 1
-    ins <- alt_width > w
-    dels <- w > alt_width
+    ins <- alt_width > w & w == 1
+    dels <- w > alt_width & alt_width == 1
     ## All should add up to the total
     if (sum(snv + ins + dels) != n)
-        stop("Some variants were unable to be uniquely identified")
+        stop("Some variants were unable to be identified as SNVs or InDels")
 
     ## Return output
     out <- character(n)

tests/testthat/test_varTypes.R

@@ -14,6 +14,5 @@ test_that("varTypes errors as expected", {
     type <- varTypes(var)
     width(var)[type == "Insertion"] <- 2
     expect_error(varTypes(var), "Some variants.+")
-    var$ALT <- 1
-    expect_error(varTypes(var), ".+is not TRUE")
+
 })

vignettes/creating_a_new_reference.Rmd

@@ -43,6 +43,14 @@ Whilst the subsequent code is demonstrated on a small genomic region, the
 complete process for creating a modified a reference can run in under 20 minutes 
 if using 4 or more cores.
 
+Importantly, it is expected that the user will have carefully prepared their set of variants such that only the following exact variant types are included.
+
+- **SNV**/**SNP**: The associated range must be of width 1 with a single base as the replacement allele
+- **Insertions**: The position being replaced must be a single nucleotide with the replacement bases being > 1nt
+- **Deletions**: The position being replaced must be > 1nt, whilst the replacement allele must be a single base
+
+Any variants which do not conform to these criteria will likely cause a series of frustrating errors when attempting to use this package.
+
 # Setup
 
 ## Installation