sync records in sampleMetadata with those in df after possible removal

microbiomeDB · Sep 12, 2024 · bbeb36e · bbeb36e
1 parent 905e3f1
commit bbeb36e
Show file tree

Hide file tree

Showing 2 changed files with 43 additions and 1 deletion.
diff --git a/R/methods-CollectionWithMetadata.R b/R/methods-CollectionWithMetadata.R
@@ -153,9 +153,10 @@ setGeneric("removeIncompleteRecords",
 #' @aliases removeIncompleteRecords,CollectionWithMetadata-method
 setMethod("removeIncompleteRecords", signature("CollectionWithMetadata"), function(object, colName = character(), verbose = c(TRUE, FALSE)) {
     verbose <- veupathUtils::matchArg(verbose)
-
     df <- getCollectionData(object, verbose = verbose)
     sampleMetadata <- getSampleMetadata(object)
+    # df may have had rows removed due to getCollectionData behavior. Subset sampleMetadata to match
+    sampleMetadata <- sampleMetadata[sampleMetadata[[object@sampleMetadata@recordIdColumn]] %in% df[[object@recordIdColumn]], ]
 
     # Remove Records with NA from data and metadata
     if (any(is.na(sampleMetadata[[colName]]))) {
@@ -171,6 +172,7 @@ setMethod("removeIncompleteRecords", signature("CollectionWithMetadata"), functi
             data = sampleMetadata,
             recordIdColumn = object@sampleMetadata@recordIdColumn
         )
+
         validObject(object)
     }
 

diff --git a/tests/testthat/test-class-CollectionWithMetadata.R b/tests/testthat/test-class-CollectionWithMetadata.R
@@ -159,4 +159,44 @@ test_that("pruneFeatures works", {
     testing <- pruneFeatures(testing, veupathUtils::predicateFactory('proportionNonZero', 0.5))
     expect_equal(nrow(testing@data), 200)
     expect_equal(ncol(testing@data), 3)
+})
+
+test_that("removeIncompleteRecords removes records from both metadata and data", {
+    nSamples <- 200
+    sampleMetadataDT <- data.table::data.table(
+        "entity.SampleID" = 1:nSamples,
+        "entity.contA" = rnorm(nSamples),
+        "entity.contB" = rnorm(nSamples),
+        "entity.contC" = rnorm(nSamples)
+    )
+    # set first 5 rows to NA
+    sampleMetadataDT[1:5, 2:ncol(sampleMetadataDT) ] <- NA
+
+    df <- data.table::data.table(
+        "entity.SampleID" = sampleMetadataDT$entity.SampleID,
+        "entity.cont1" = rnorm(nSamples),
+        "entity.cont2" = rnorm(nSamples),
+        "entity.cont3" = rnorm(nSamples)
+    )
+    # Set second 5 rows to NA
+    df[6:10, 2:ncol(df) ] <- NA
+
+    sampleMetadata <- SampleMetadata(
+        data = sampleMetadataDT,
+        recordIdColumn = 'entity.SampleID'
+    )
+
+    test_collection <- CollectionWithMetadata(
+        name = 'test',
+        data = df,
+        sampleMetadata = sampleMetadata,
+        recordIdColumn = 'entity.SampleID'
+    )
+
+    result <- removeIncompleteRecords(test_collection, 'entity.contA', verbose=F)
+    expect_equal(nrow(result@data), nSamples-10)
+    expect_equal(ncol(result@data), 4)
+    expect_equal(nrow(result@sampleMetadata@data), nSamples-10)
+    expect_equal(ncol(result@sampleMetadata@data), 4)
+
 })