From 71233261d20b566c86a539101a77281ee16508df Mon Sep 17 00:00:00 2001
From: "Vera N. Karlbauer" <vera@karlbauer.org>
Date: Tue, 13 Aug 2024 18:05:00 +0200
Subject: [PATCH] Replaced all RGSets with gRatioSets
---
scripts/6_filtering_cpgs.Rmd | 37 +++++++++++++++++++-----------------
1 file changed, 20 insertions(+), 17 deletions(-)
diff --git a/scripts/6_filtering_cpgs.Rmd b/scripts/6_filtering_cpgs.Rmd
index e0870b2..db4c677 100644
--- a/scripts/6_filtering_cpgs.Rmd
+++ b/scripts/6_filtering_cpgs.Rmd
@@ -105,9 +105,10 @@ if (user_choices$array_type == "v2") {
filter(CpG_name %in% replicates$CpG_name) %>%
filter(!(IlmnID %in% keep_replicates$IlmnID))
- # exclude replicates from betas, RGSet, detP and save data
- RGSet_clean <- subsetByLoci(RGSet_clean, excludeLoci = exclude_replicates$IlmnID)
- save(RGSet_clean, file = paste0(user_choices$project_name, "/processed_data/RGSet_clean.Rdata"))
+ # exclude replicates from betas, gRatioSet, detP and save data
+ keep_gratioset <- !(featureNames(gRatioSet_clean) %in% exclude_replicates$IlmnID)
+ gRatioSet_clean <- gRatioSet_clean[keep_gratioset,]
+ save(gRatioSet_clean, file = paste0(user_choices$project_name, "/processed_data/gRatioSet_clean.Rdata"))
keep_betas <- !(rownames(Betas_clean) %in% exclude_replicates$IlmnID)
Betas_clean <- Betas_clean[keep_betas,]
@@ -126,13 +127,13 @@ if (user_choices$array_type == "v2") {
keep_betas_df <- as.data.frame(keep_betas)
cat(paste0(nrow(exclude_replicates), " replicate probes were removed"))
- dim_RGSet_filtered <- dim(RGSet_clean)
+ dim_gRatioSet_filtered <- dim(gRatioSet_clean)
dim_Betas_filtered <- dim(Betas_clean)
step_number <- c("4", "4")
step <- c("Filter replicates", "Filter replicates")
- data_class <- c("RGSet", "Betas")
- samples <- c(dim_RGSet_filtered[2], dim_Betas_filtered[2])
- probes <- c(dim_RGSet_filtered[1], dim_Betas_filtered[1])
+ data_class <- c("gRatioSet", "Betas")
+ samples <- c(dim_gRatioSet_filtered[2], dim_Betas_filtered[2])
+ probes <- c(dim_gRatioSet_filtered[1], dim_Betas_filtered[1])
table_preprocessing_adding <- data.frame(step_number, step, data_class, samples, probes)
summary_table_preprocessing <- bind_rows(summary_table_preprocessing, table_preprocessing_adding)
@@ -371,7 +372,8 @@ if (user_choices$array_type == "v2") {
keep_betas <- !(rownames(Betas_clean_filtered) %in% v2_mapping_inacc$IlmnID)
Betas_clean_filtered <- Betas_clean_filtered[keep_betas,]
- RGSet_clean_filtered <- subsetByLoci(RGSet_clean_filtered, excludeLoci = v2_mapping_inacc$IlmnID)
+ keep_gratioset <- !(featureNames(gRatioSet_clean_filtered) %in% v2_mapping_inacc$IlmnID)
+ gRatioSet_clean_filtered <- gRatioSet_clean_filtered[keep_gratioset,]
}
```
@@ -381,13 +383,13 @@ if (user_choices$array_type == "v2") {
cat(paste0(keep_betas_df %>% filter(keep_betas == FALSE) %>% nrow(),
" CpGs show known mapping inaccuracies"), sep = "<br>\n")
- dim_RGSet_filtered <- dim(RGSet_clean_filtered)
+ dim_gRatioSet_filtered <- dim(gRatioSet_clean_filtered)
dim_Betas_filtered <- dim(Betas_clean_filtered)
step_number <- c("12", "12")
step <- c("Filter Mapping Inaccuracies", "Filter Mapping Inaccuracies")
- data_class <- c("RGSet", "Betas")
- samples <- c(dim_RGSet_filtered[2], dim_Betas_filtered[2])
- probes <- c(dim_RGSet_filtered[1], dim_Betas_filtered[1])
+ data_class <- c("gRatioSet", "Betas")
+ samples <- c(dim_gRatioSet_filtered[2], dim_Betas_filtered[2])
+ probes <- c(dim_gRatioSet_filtered[1], dim_Betas_filtered[1])
table_preprocessing_adding <- data.frame(step_number, step, data_class, samples, probes)
summary_table_preprocessing <- bind_rows(summary_table_preprocessing, table_preprocessing_adding)
@@ -401,7 +403,8 @@ if (user_choices$array_type == "v2") {
keep_betas <- !(rownames(Betas_clean_filtered) %in% v2_flagged_probes$IlmnID)
Betas_clean_filtered <- Betas_clean_filtered[keep_betas,]
- RGSet_clean_filtered <- subsetByLoci(RGSet_clean_filtered, excludeLoci = v2_flagged_probes$IlmnID)
+ keep_gratioset <- !(featureNames(gRatioSet_clean_filtered) %in% v2_flagged_probes$IlmnID)
+ gRatioSet_clean_filtered <- gRatioSet_clean_filtered[keep_gratioset,]
}
```
@@ -411,13 +414,13 @@ if (user_choices$array_type == "v2") {
cat(paste0(keep_betas_df %>% filter(keep_betas == FALSE) %>% nrow(),
" CpGs were flagged by Illumina"), sep = "<br>\n")
- dim_RGSet_filtered <- dim(RGSet_clean_filtered)
+ dim_gRatioSet_filtered <- dim(gRatioSet_clean_filtered)
dim_Betas_filtered <- dim(Betas_clean_filtered)
step_number <- c("13", "13")
step <- c("Filter Flagged Probes", "Filter Flagged Probes")
- data_class <- c("RGSet", "Betas")
- samples <- c(dim_RGSet_filtered[2], dim_Betas_filtered[2])
- probes <- c(dim_RGSet_filtered[1], dim_Betas_filtered[1])
+ data_class <- c("gRatioSet", "Betas")
+ samples <- c(dim_gRatioSet_filtered[2], dim_Betas_filtered[2])
+ probes <- c(dim_gRatioSet_filtered[1], dim_Betas_filtered[1])
table_preprocessing_adding <- data.frame(step_number, step, data_class, samples, probes)
summary_table_preprocessing <- bind_rows(summary_table_preprocessing, table_preprocessing_adding)