fast conversion of cov_table dataset (#63)

* fast conversion of cov_table dataset * fast conversion of cov_table dataset in EXOMEDEPTHCOV and CODEXCOV packages
ZSI-Bio · Dec 15, 2017 · c0bf3f5 · c0bf3f5
1 parent c2f23f6
commit c0bf3f5
Show file tree

Hide file tree

Showing 5 changed files with 23 additions and 26 deletions.
diff --git a/R/CODEXCOV/R/functions_CODEXCOV.R b/R/CODEXCOV/R/functions_CODEXCOV.R
@@ -10,14 +10,14 @@ library(CODEX)
 #' @examples
 #' coverageObj1
 coverageObj1 <- function(cov_table, sampname, targets_for_chr, chr){
-  Y <- matrix(data=as.integer(0), nrow = nrow(targets_for_chr), ncol = length(sampname))
+  Y <- matrix(data=as.integer(0), nrow = nrow(targets_for_chr), ncol = 0)
+  for(sample in sampname) {
+    cov_targets_for_sample <- cov_table[cov_table[,"sample_name"] == sample,]
+    cov_targets_for_sample <- cov_targets_for_sample[with(cov_targets_for_sample, order(target_id)), ]
+    Y <- cbind(Y, cov_targets_for_sample[,"read_count"])
+  }
   colnames(Y) <- sampname
   rownames(Y) <- targets_for_chr[,"target_id"]
-  cov_targets_for_chr <- cov_table[cov_table[,"chr"] == chr,]
-  for(i in 1:nrow(cov_targets_for_chr)) {
-    cov_row <- cov_targets_for_chr[i,]
-    Y[toString(cov_row[,"target_id"]),toString(cov_row[,"sample_name"])] = as.integer(cov_row[,"read_count"])
-  }
   return(list(Y=Y))
 }
 

diff --git a/R/EXOMEDEPTHCOV/R/functions_EXOMEDEPTHCOV.R b/R/EXOMEDEPTHCOV/R/functions_EXOMEDEPTHCOV.R
@@ -1,14 +1,14 @@
 library(ExomeDepth)
 
 coverageObj1 <- function(cov_table, sampname, targets_for_chr, chr){
-  Y <- matrix(data=as.integer(0), nrow = nrow(targets_for_chr), ncol = length(sampname))
+  Y <- matrix(data=as.integer(0), nrow = nrow(targets_for_chr), ncol = 0)
+  for(sample in sampname) {
+    cov_targets_for_sample <- cov_table[cov_table[,"sample_name"] == sample,]
+    cov_targets_for_sample <- cov_targets_for_sample[with(cov_targets_for_sample, order(target_id)), ]
+    Y <- cbind(Y, cov_targets_for_sample[,"read_count"])
+  }
   colnames(Y) <- sampname
   rownames(Y) <- targets_for_chr[,"target_id"]
-  cov_targets_for_chr <- cov_table[cov_table[,"chr"] == chr,]
-  for(i in 1:nrow(cov_targets_for_chr)) {
-    cov_row <- cov_targets_for_chr[i,]
-    Y[toString(cov_row[,"target_id"]),toString(cov_row[,"sample_name"])] = as.integer(cov_row[,"read_count"])
-  }
   return(list(Y=Y))
 }
 

diff --git a/R/TARGET.QC/R/functions_TARGET.QC.R b/R/TARGET.QC/R/functions_TARGET.QC.R
@@ -1,14 +1,14 @@
 library(CODEX)
 
 coverageObj1 <- function(cov_table, sampname, targets_for_chr, chr){
-  Y <- matrix(data=as.integer(0), nrow = nrow(targets_for_chr), ncol = length(sampname))
+  Y <- matrix(data=as.integer(0), nrow = nrow(targets_for_chr), ncol = 0)
+  for(sample in sampname) {
+    cov_targets_for_sample <- cov_table[cov_table[,"sample_name"] == sample,]
+    cov_targets_for_sample <- cov_targets_for_sample[with(cov_targets_for_sample, order(target_id)), ]
+    Y <- cbind(Y, cov_targets_for_sample[,"read_count"])
+  }
   colnames(Y) <- sampname
   rownames(Y) <- targets_for_chr[,"target_id"]
-  cov_targets_for_chr <- cov_table[cov_table[,"chr"] == chr,]
-  for(i in 1:nrow(cov_targets_for_chr)) {
-    cov_row <- cov_targets_for_chr[i,]
-    Y[toString(cov_row[,"target_id"]),toString(cov_row[,"sample_name"])] = as.integer(cov_row[,"read_count"])
-  }
   return(list(Y=Y))
 }
 

diff --git a/R/TARGET.QC/R/run_TARGET.QC.R b/R/TARGET.QC/R/run_TARGET.QC.R
@@ -16,7 +16,6 @@ run_TARGET.QC <- function(mapp_thresh,
   #K_from <- 1
   #K_to <- 9
   #lmax <- 200
-
   sampname <- unique(cov_table[,"sample_name"])
   targets <- cov_table[,c("target_id", "chr", "pos_min", "pos_max")]
   targets <- targets[!duplicated(targets[,"target_id"]),]
@@ -32,7 +31,6 @@ run_TARGET.QC <- function(mapp_thresh,
       next()
     }
     Y <- coverageObj1(cov_table, sampname, targets_for_chr, chr)$Y
-
     gcmapp1_result <- gcmapp1(chr, ref)
     gc <- gcmapp1_result$gc
     mapp <- gcmapp1_result$mapp
@@ -43,11 +41,10 @@ run_TARGET.QC <- function(mapp_thresh,
     Y_qc <- qcObj1_result$Y_qc
     sampname_qc <- qcObj1_result$sampname_qc
     ref_qc <- qcObj1_result$ref_qc
-    for(i in 1:nrow(Y_qc)) {
-      for (j in 1:ncol(Y_qc)) {
-        new_cov_table_qc_row <- c(colnames(Y_qc)[j], rownames(Y_qc)[i], chr, start(ref_qc)[i], end(ref_qc)[i], Y_qc[i,j])
-        cov_table_qc <- rbind(cov_table_qc, new_cov_table_qc_row)
-      }
+    colnames(Y_qc) <- sampname_qc
+    for(sample in colnames(Y_qc)) {
+      new_cov_table_qc_rows <- cbind(sample, rownames(Y_qc), chr, start(ref_qc), end(ref_qc), Y_qc[,sample])
+      cov_table_qc <- rbind(cov_table_qc, new_cov_table_qc_rows)
     }
   }
   cov_table_qc <- as.data.frame(cov_table_qc)

diff --git a/R/tests/test_TARGET.QC_wrapper.R b/R/tests/test_TARGET.QC_wrapper.R
@@ -77,5 +77,5 @@ test_that("basic test for run_wrapper_TARGET.QC function",{
                                      gc_thresh_to,
                                      cov_table)
   expect_equal(ncol(cov_table), 6)
-  expect_equal(nrow(cov_table), 18)
+  expect_equal(nrow(cov_table), 24)
 })