Create PCL_CCS_merge.R

Add PCL_CCS_merge.R: Merge CCS values to PubChemLite exposomics file

pubchemlite/R/PCL_CCS_merge.R

+## Copyright (C) 2020 by University of Luxembourg
+
+
+## Rscript : PCL_CCS_merge.R
+## Add CCS to PubChemLite exposomics file.
+## Input : pubchemlite_ccs.csv, Pubchemlite_exposomics.csv
+
+pcl <- read.csv("PubChemLite_exposomics_20240329.csv", check.names = FALSE)
+ccs <- read.csv("pubchemlite_CCS_20240410.csv", check.names = FALSE)
+
+# Define the columns to be extracted from ccs
+#ccs_columns <- ccs[,c(1,5:12)]  # Columns 5 to 12 in ccs
+
+# Create a lookup table from ccs with "cid" and selected columns
+lookup_table <- ccs[,c(1,5:12)]
+
+# Merge the lookup table with pcl based on "identifier" and "cid"
+merged_data <- merge(pcl, lookup_table, by.x = "Identifier", by.y = "cid", all.x = TRUE)
+
+# Define new column names corresponding to the selected columns from ccs
+new_column_names <- c(
+  "pred_CCS_A2_[M+H]+",
+  "pred_CCS_A2_[M+Na]+",
+  "pred_CCS_A2_[M+NH4]+",
+  "pred_CCS_A2_[M+K]+",
+  "pred_CCS_A2_[M-H]-",
+  "pred_CCS_A2_[M+Na-2H]-",
+  "pred_CCS_A2_[M]+",
+  "pred_CCS_A2_[M]-"
+)
+
+# Assign new column names to the extracted columns in merged_data
+colnames(merged_data)[(ncol(merged_data) - length(new_column_names) + 1):ncol(merged_data)] <- new_column_names
+
+# Write the updated data to a new CSV file
+write.csv(merged_data, "PubChemLite_CCSbase_20240329.csv", row.names = FALSE)