Check dataset spreadsheets for duplicate sample/replicate/locus combinations

NEWS.md

@@ -1,8 +1,11 @@
 # chiimp dev
 
+ * Made `load_dataset` warn about repeated Sample+Replicate+Locus entries
+   across rows ([#90])
  * Made `load_config` warn about any unrecognized configuration file entries
    ([#88])
 
+[#90]: https://github.com/ShawHahnLab/chiimp/pull/90
 [#88]: https://github.com/ShawHahnLab/chiimp/pull/88
 
 # chiimp 0.4.0

R/io.R

@@ -252,6 +252,14 @@ load_dataset <- function(fp, ...) {
     warning(paste("Missing columns in dataset table:",
                   dataset_cols[col.missing]))
   }
+  # check for duplicated Sample+Replicate+Locus entries
+  cts <- table(with(data, paste(Sample, Replicate, Locus, sep = "/")))
+  dups <- names(cts[cts > 1])
+  if (length(dups) > 0) {
+    warning(paste(
+      "Duplicated sample/replicate/locus entries for",
+      paste(dups, collapse = ", ")))
+  }
   rownames(data) <- make_rownames(data)
   data
 }

R/util.R

@@ -201,9 +201,4 @@ logmsg <- function(msg, col2=as.character(Sys.time()), end="\n") {
   }
   # stderr: file descriptor 2
   cat(paste0(msg, end), file = 2)
-}
-
-# append an empty string to each the given files
-touch <- function(fps) {
-  lapply(fps, function(fp) cat("", file = fp, append = TRUE))
-}
+}

tests/testthat/data/io/dataset.csv

@@ -0,0 +1,61 @@
+Filename,Replicate,Sample,Locus
+1-1-1.fasta,1,1,1
+2-1-1.fasta,2,1,1
+3-1-1.fasta,3,1,1
+1-2-1.fasta,1,2,1
+2-2-1.fasta,2,2,1
+3-2-1.fasta,3,2,1
+1-3-1.fasta,1,3,1
+2-3-1.fasta,2,3,1
+3-3-1.fasta,3,3,1
+1-4-1.fasta,1,4,1
+2-4-1.fasta,2,4,1
+3-4-1.fasta,3,4,1
+1-5-1.fasta,1,5,1
+2-5-1.fasta,2,5,1
+3-5-1.fasta,3,5,1
+1-1-2.fasta,1,1,2
+2-1-2.fasta,2,1,2
+3-1-2.fasta,3,1,2
+1-2-2.fasta,1,2,2
+2-2-2.fasta,2,2,2
+3-2-2.fasta,3,2,2
+1-3-2.fasta,1,3,2
+2-3-2.fasta,2,3,2
+3-3-2.fasta,3,3,2
+1-4-2.fasta,1,4,2
+2-4-2.fasta,2,4,2
+3-4-2.fasta,3,4,2
+1-5-2.fasta,1,5,2
+2-5-2.fasta,2,5,2
+3-5-2.fasta,3,5,2
+1-1-A.fasta,1,1,A
+2-1-A.fasta,2,1,A
+3-1-A.fasta,3,1,A
+1-2-A.fasta,1,2,A
+2-2-A.fasta,2,2,A
+3-2-A.fasta,3,2,A
+1-3-A.fasta,1,3,A
+2-3-A.fasta,2,3,A
+3-3-A.fasta,3,3,A
+1-4-A.fasta,1,4,A
+2-4-A.fasta,2,4,A
+3-4-A.fasta,3,4,A
+1-5-A.fasta,1,5,A
+2-5-A.fasta,2,5,A
+3-5-A.fasta,3,5,A
+1-1-B.fasta,1,1,B
+2-1-B.fasta,2,1,B
+3-1-B.fasta,3,1,B
+1-2-B.fasta,1,2,B
+2-2-B.fasta,2,2,B
+3-2-B.fasta,3,2,B
+1-3-B.fasta,1,3,B
+2-3-B.fasta,2,3,B
+3-3-B.fasta,3,3,B
+1-4-B.fasta,1,4,B
+2-4-B.fasta,2,4,B
+3-4-B.fasta,3,4,B
+1-5-B.fasta,1,5,B
+2-5-B.fasta,2,5,B
+3-5-B.fasta,3,5,B

tests/testthat/data/io/dataset_dups.csv

@@ -0,0 +1,65 @@
+Filename,Replicate,Sample,Locus
+1-1-1.fasta,1,1,1
+2-1-1.fasta,2,1,1
+3-1-1.fasta,3,1,1
+1-2-1.fasta,1,2,1
+2-2-1.fasta,2,2,1
+3-2-1.fasta,3,2,1
+1-3-1.fasta,1,3,1
+2-3-1.fasta,2,3,1
+3-3-1.fasta,3,3,1
+1-4-1.fasta,1,4,1
+2-4-1.fasta,2,4,1
+3-4-1.fasta,3,4,1
+1-5-1.fasta,1,5,1
+2-5-1.fasta,2,5,1
+3-5-1.fasta,3,5,1
+1-1-2.fasta,1,1,2
+2-1-2.fasta,2,1,2
+3-1-2.fasta,3,1,2
+1-2-2.fasta,1,2,2
+2-2-2.fasta,2,2,2
+3-2-2.fasta,3,2,2
+1-3-2.fasta,1,3,2
+2-3-2.fasta,2,3,2
+3-3-2.fasta,3,3,2
+1-4-2.fasta,1,4,2
+2-4-2.fasta,2,4,2
+3-4-2.fasta,3,4,2
+1-5-2.fasta,1,5,2
+2-5-2.fasta,2,5,2
+3-5-2.fasta,3,5,2
+1-1-A.fasta,1,1,A
+2-1-A.fasta,2,1,A
+3-1-A.fasta,3,1,A
+1-2-A.fasta,1,2,A
+2-2-A.fasta,2,2,A
+3-2-A.fasta,3,2,A
+1-3-A.fasta,1,3,A
+2-3-A.fasta,2,3,A
+3-3-A.fasta,3,3,A
+1-4-A.fasta,1,4,A
+2-4-A.fasta,2,4,A
+3-4-A.fasta,3,4,A
+1-5-A.fasta,1,5,A
+2-5-A.fasta,2,5,A
+3-5-A.fasta,3,5,A
+1-1-B.fasta,1,1,B
+2-1-B.fasta,2,1,B
+3-1-B.fasta,3,1,B
+1-2-B.fasta,1,2,B
+2-2-B.fasta,2,2,B
+3-2-B.fasta,3,2,B
+1-3-B.fasta,1,3,B
+2-3-B.fasta,2,3,B
+3-3-B.fasta,3,3,B
+1-4-B.fasta,1,4,B
+2-4-B.fasta,2,4,B
+3-4-B.fasta,3,4,B
+1-5-B.fasta,1,5,B
+2-5-B.fasta,2,5,B
+3-5-B.fasta,3,5,B
+1-1-1-alt.fasta,1,1,1
+1-1-2-alt.fasta,1,1,2
+1-1-A-alt.fasta,1,1,A
+1-1-B-alt.fasta,1,1,B

tests/testthat/data/io/locus_attrs.csv

@@ -0,0 +1,5 @@
+Locus,LengthMin,LengthMax,LengthBuffer,Motif,Primer,ReversePrimer
+A,131,179,20,TAGA,TATCACTGGTGTTAGTCCTCTG,CACAGTTGTGTGAGCCAGTC
+B,194,235,20,TAGA,AGTCTCTCTTTCTCCTTGCA,TAGGAGCCTGTGGTCCTGTT
+1,232,270,20,TATC,ACAGTCAAGAATAACTGCCC,CTGTGGCTCAAAAGCTGAAT
+2,218,337,20,TCCA,TTGTCTCCCCAGTTGCTA,TCTGTCATAAACCGTCTGCA

tests/testthat/data/io/locus_attrs_dups.csv

@@ -0,0 +1,5 @@
+Locus,LengthMin,LengthMax,LengthBuffer,Motif,Primer,ReversePrimer
+A,131,179,20,TAGA,TATCACTGGTGTTAGTCCTCTG,CACAGTTGTGTGAGCCAGTC
+A,194,235,20,TAGA,AGTCTCTCTTTCTCCTTGCA,TAGGAGCCTGTGGTCCTGTT
+1,232,270,20,TATC,ACAGTCAAGAATAACTGCCC,CTGTGGCTCAAAAGCTGAAT
+2,218,337,20,TCCA,TTGTCTCCCCAGTTGCTA,TCTGTCATAAACCGTCTGCA

tests/testthat/data/io/locus_attrs_wrongcol.csv

@@ -0,0 +1,5 @@
+Locus,length_min,LengthMax,LengthBuffer,Motif,Primer,ReversePrimer
+A,131,179,20,TAGA,TATCACTGGTGTTAGTCCTCTG,CACAGTTGTGTGAGCCAGTC
+B,194,235,20,TAGA,AGTCTCTCTTTCTCCTTGCA,TAGGAGCCTGTGGTCCTGTT
+1,232,270,20,TATC,ACAGTCAAGAATAACTGCCC,CTGTGGCTCAAAAGCTGAAT
+2,218,337,20,TCCA,TTGTCTCCCCAGTTGCTA,TCTGTCATAAACCGTCTGCA

tests/testthat/data/io/misc.csv

@@ -0,0 +1,3 @@
+Vec1,Vec2,Vec3
+A,B,C
+D,E,F

tests/testthat/helper.R

@@ -0,0 +1,17 @@
+# execute expression inside of a temporary directory, and remove the directory
+# afterward
+within_tmpdir <- function(expr) {
+  here <- getwd()
+  data.dir <- tempfile()
+  dir.create(data.dir)
+  setwd(data.dir)
+  tryCatch(eval(expr), finally = {
+    unlink(x = data.dir, recursive = TRUE)
+    setwd(here)
+  })
+}
+
+# append an empty string to each the given files
+touch <- function(fps) {
+  lapply(fps, function(fp) cat("", file = fp, append = TRUE))
+}