Merge pull request #450 from Qile0317/tcrfix

Minor `combineTCR` fix & refactor `combineBCR`
BorchLab · Nov 26, 2024 · 7d6ea5e · 7d6ea5e
2 parents b945dc1 + 9811879
commit 7d6ea5e
Show file tree

Hide file tree

Showing 7 changed files with 157 additions and 155 deletions.
diff --git a/R/combineContigs.R b/R/combineContigs.R
diff --git a/R/combineExpression.R b/R/combineExpression.R
@@ -223,5 +223,5 @@ getHighBarcodeMismatchWarning <- function() paste(
     "< 1% of barcodes match: Ensure the barcodes in the single-cell object",
     "match the barcodes in the combined immune receptor output from",
     "scRepertoire. If getting this error, please check",
-    "https://www.borch.dev/uploads/screpertoire/articles/faq."
+    "https://www.borch.dev/uploads/screpertoire/articles/faq"
 )
diff --git a/R/typecheck.R b/R/typecheck.R
@@ -1,8 +1,14 @@
 # base R type check functions
 
+isNonEmptyDataFrame <- function(obj) {
+    is.data.frame(obj) && sum(dim(obj)) > 0
+}
+assertthat::on_failure(isNonEmptyDataFrame) <- function(call, env) {
+    paste0(deparse(call$obj), " is not a non-empty `data.frame`")
+}
+
 isListOfNonEmptyDataFrames <- function(obj) {
-    is.list(obj) &&
-        all(sapply(obj, function(x) is.data.frame(x) && sum(dim(x)) > 0))
+    is.list(obj) && all(sapply(obj, isNonEmptyDataFrame))
 }
 assertthat::on_failure(isListOfNonEmptyDataFrames) <- function(call, env) {
     paste0(deparse(call$obj), " is not a list of non-empty `data.frame`s")

diff --git a/R/utils.R b/R/utils.R
@@ -371,8 +371,8 @@
 #' @keywords internal
 .constructConDfAndParseTCR <- function(data2) {
   rcppConstructConDfAndParseTCR(
-    data2 %>% dplyr::arrange(., chain, cdr3_nt),
-    unique(data2[[1]]) # 1 is the index of the barcode column
+    dplyr::arrange(data2, chain, cdr3_nt),
+    uniqueData2Barcodes = unique(data2$barcode)
   )
 }
 
@@ -383,7 +383,7 @@
 .parseBCR <- function (Con.df, unique_df, data2) {
   barcodeIndex <- rcppConstructBarcodeIndex(unique_df, data2$barcode)
   for (y in seq_along(unique_df)) {
-    location.i <- barcodeIndex[[y]] # *may* be wrong but should be fine. Test on old version first
+    location.i <- barcodeIndex[[y]]
 
     for (z in seq_along(location.i)) {
       where.chain <- data2[location.i[z],"chain"]

diff --git a/man/combineBCR.Rd b/man/combineBCR.Rd
diff --git a/man/combineTCR.Rd b/man/combineTCR.Rd
diff --git a/src/constructConDfAndparseTCR.cpp b/src/constructConDfAndparseTCR.cpp
@@ -43,7 +43,7 @@ class TcrParser {
     TcrParser(
         Rcpp::DataFrame& data2, std::vector<std::string>& uniqueData2Barcodes
     ) {
-        // construct conDf, initializaing the matrix to "NA" *strings*
+        // construct conDf, initializing the matrix to "NA" *strings*
         conDf = scRepHelper::initStringMatrix(
             7, uniqueData2Barcodes.size(), "NA"
         );
@@ -60,12 +60,12 @@ class TcrParser {
 
         // construct barcodeIndex
         barcodeIndex = constructBarcodeIndex(
-            uniqueData2Barcodes, Rcpp::as<std::vector<std::string>>(data2[data2.findName("barcode")])
+            uniqueData2Barcodes, data2[data2.findName("barcode")]
         );
     }
 
     // Rcpp implementation of .parseTCR()
-    void parseTCR() {
+    TcrParser& parseTCR() {
         for (int y = 0; y < (int) conDf[0].size(); y++) {
             for (int index : barcodeIndex[y]) {
                 std::string chainType = std::string(data2ChainTypes[index]);
@@ -78,6 +78,7 @@ class TcrParser {
                 }
             }
         }
+        return *this;
     }
 
     // parseTCR() helpers
@@ -122,7 +123,5 @@ class TcrParser {
 Rcpp::DataFrame rcppConstructConDfAndParseTCR(
     Rcpp::DataFrame& data2, std::vector<std::string> uniqueData2Barcodes
 ) {
-    TcrParser parser = TcrParser(data2, uniqueData2Barcodes);
-    parser.parseTCR();
-    return parser.getConDf();
+    return TcrParser(data2, uniqueData2Barcodes).parseTCR().getConDf();
 }