Only perform replicate removal for EPICv2 array

scripts/6_filtering_cpgs.Rmd

@@ -69,67 +69,74 @@ annotations_clean = getAnnotation(RGSet_clean)
 ### Detect replicates (v2 only)
 
 ```{r, detect replicates}
-# add non-unique CpG name
-CpG_names <- as.data.frame(rownames(Betas_clean))
-colnames(CpG_names) <- c("IlmnID")
-CpG_names$CpG_name <- str_sub(CpG_names$IlmnID, 1, 10)
-# isolate replicate IDs
-replicates <- CpG_names %>%
-  group_by(CpG_name) %>%
-  filter(n()>1) %>% 
-  distinct(CpG_name)
-
-# link to detP values
-CpG_names_detP <- left_join(CpG_names, as.data.frame(cbind(rownames(detP_clean), rowMeans(detP_clean))), by = join_by("IlmnID" == "V1"))
-colnames(CpG_names_detP) <- c("IlmnID", "CpG_name", "detP_value")
-
-cat(paste0(nrow(replicates), " replicate CpGs were detected"))
+if (user_choices$array_type == "v2") {
+  # add non-unique CpG name
+  CpG_names <- as.data.frame(rownames(Betas_clean))
+  colnames(CpG_names) <- c("IlmnID")
+  CpG_names$CpG_name <- str_sub(CpG_names$IlmnID, 1, 10)
+  # isolate replicate IDs
+  replicates <- CpG_names %>%
+    group_by(CpG_name) %>%
+    filter(n()>1) %>% 
+    distinct(CpG_name)
+  
+  # link to detP values
+  CpG_names_detP <- left_join(CpG_names, as.data.frame(cbind(rownames(detP_clean), 
+                                                             rowMeans(detP_clean))), by = join_by("IlmnID" == "V1"))
+  colnames(CpG_names_detP) <- c("IlmnID", "CpG_name", "detP_value")
+  
+  cat(paste0(nrow(replicates), " replicate CpGs were detected"))
+  }
 ```
 
 ```{r, exclude replicates and save data, include = FALSE}
-## create list of replicates with lowest detP values, if detPs are equal, take first replicate
-keep_replicates <- CpG_names_detP %>%
-  filter(CpG_name %in% replicates$CpG_name) %>%
-  group_by(CpG_name) %>%
-  slice_min(detP_value) %>%
-  ungroup() 
-keep_replicates <- keep_replicates %>%
-  filter(duplicated(CpG_name) == FALSE)
-
-# create list of replicates to exclude (CpGs in replicate list but probe not in keep list)
-exclude_replicates <- CpG_names_detP %>%
-  filter(CpG_name %in% replicates$CpG_name) %>%
-  filter(!(IlmnID %in% keep_replicates$IlmnID))
+if (user_choices$array_type == "v2") {
+  # create list of replicates with lowest detP values, if detPs are equal, take first replicate
+  keep_replicates <- CpG_names_detP %>%
+    filter(CpG_name %in% replicates$CpG_name) %>%
+    group_by(CpG_name) %>%
+    slice_min(detP_value) %>%
+    ungroup() 
+  keep_replicates <- keep_replicates %>%
+    filter(duplicated(CpG_name) == FALSE)
   
-# 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"))
-
-keep_betas <- !(rownames(Betas_clean) %in% exclude_replicates$IlmnID) 
-Betas_clean <- Betas_clean[keep_betas,]
-save(Betas_clean, file = paste0(user_choices$project_name, "/processed_data/Betas_clean.Rdata"))
-
-keep_detP <- !(rownames(detP_clean) %in% exclude_replicates$IlmnID) 
-detP_clean <- detP_clean[keep_detP,]
-save(detP_clean, file = paste0(user_choices$project_name, "/processed_data/detP_clean.Rdata"))
+  # create list of replicates to exclude (CpGs in replicate list but probe not in keep list)
+  exclude_replicates <- CpG_names_detP %>%
+    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"))
+  
+  keep_betas <- !(rownames(Betas_clean) %in% exclude_replicates$IlmnID) 
+  Betas_clean <- Betas_clean[keep_betas,]
+  save(Betas_clean, file = paste0(user_choices$project_name, "/processed_data/Betas_clean.Rdata"))
+  
+  keep_detP <- !(rownames(detP_clean) %in% exclude_replicates$IlmnID) 
+  detP_clean <- detP_clean[keep_detP,]
+  save(detP_clean, file = paste0(user_choices$project_name, "/processed_data/detP_clean.Rdata"))
+  }
 ```
 
 ### Exclude replicates 
 
 ```{r, info output for replicate removal, results='asis'}
-keep_betas_df <- as.data.frame(keep_betas)
-cat(paste0(nrow(exclude_replicates), " replicate probes were removed"))
-
-dim_RGSet_filtered <- dim(RGSet_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])
+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_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])
 
-table_preprocessing_adding <- data.frame(step_number, step, data_class, samples, probes)
-summary_table_preprocessing <- bind_rows(summary_table_preprocessing, table_preprocessing_adding)
+  table_preprocessing_adding <- data.frame(step_number, step, data_class, samples, probes)
+  summary_table_preprocessing <- bind_rows(summary_table_preprocessing, table_preprocessing_adding)
+  }
 ```
 
 ## Removal of failed probes in one or more samples