Accessing actual content of XML file in R? - r

Question

I am working with a well-structured XML file with the following initial content:
<?xml version="1.0" encoding="UTF-8"?>
<drugbank xmlns="http://www.drugbank.ca" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.drugbank.ca http://www.drugbank.ca/docs/drugbank.xsd" version="5.0" exported-on="2017-07-06">
<drug type="biotech" created="2005-06-13" updated="2016-08-17">
<drugbank-id primary="true">DB00001</drugbank-id>
<drugbank-id>BTD00024</drugbank-id>
<drugbank-id>BIOD00024</drugbank-id>
<name>Lepirudin</name>
<description>Lepirudin is identical to natural hirudin except for substitution of leucine for isoleucine at the N-terminal end of the molecule and the absence of a sulfate group on the tyrosine at position 63. It is produced via yeast cells. Bayer ceased the production of lepirudin (Refludan) effective May 31, 2012.</description>
<cas-number>138068-37-8</cas-number>
<unii>Y43GF64R34</unii>
<state>liquid</state>
<groups>
<group>approved</group>
</groups>
...
This file consists of many nodes, each representing one drug. I am aiming to extract two specific fields from each node of this file: name and drugbank-id primary="true"
... and save these to a neatly formatted table (with one column for name and the second column for drugbank-id).
I have reviewed a number of tutorials and had success with accessing the higher levels of this XML table structure, but where the examples provide syntax to access the actual values (e.g. specific drug names), this code is not working for me.
This is my current code:
library(XML)
# Save the database file as a tree structure
xmldata = xmlRoot(xmlTreeParse("DrugBank_TruncatedDatabase_v3_Small.xml"))
# Number of nodes in the entire database file
NumNodes <- xmlSize(xmldata)
# Create array structure to hold DrugBank ID values
DB_ID <- array(1:NumNodes, dim=c(1,NumNodes,1))
# Create array structure to hold Drug Name values
DrugName <- array(1:NumNodes, dim=c(1,NumNodes,1))
# for each node (i.e. each drug) in the database
for (i in 1:NumNodes){
# Assign the Drug Names to easy-to-comprehend DrugName array
DrugName[i] <- xmldata[[i]][["name"]]
# Assign the DrugBank ID numbers to easy-to-comprehend DB_ID array
DB_ID[i] <- xmldata[[i]][["drugbank-id"]]
}
EdgeListTable = data.frame(DrugName, DB_ID)
write.table(EdgeListTable, file="Output1.txt", quote=F)
The output file contains the following text, which is a level higher than I want:
X.name. X.name..1 X.name..2 X.name..3 X.drugbank.id. X.drugbank.id..1 X.drugbank.id..2 X.drugbank.id..3
1 name name name name drugbank-id drugbank-id drugbank-id drugbank-id
If I try:
xmlSApply(xmldata, function(x) xmlSApply(x, xmlValue))
...my output looks like:
$drug
$drug$drugbank-id
[1] "DB00001"
$drug$drugbank-id
[1] "BTD00024"
$drug$drugbank-id
[1] "BIOD00024"
$drug$name
[1] "Lepirudin"
...
...but after experimentation, I'm not sure how to actually access the values needed.
I appreciate any advice regarding the best way to store the values in the two fields of interest as a table.
============================================================
Update: I am able to extract the desired values using the following code:
DrugBankData <- xmlSApply(xmldata, function(x) xmlSApply(x, xmlValue))
for (i in 1:NumNodes){
DB_ID[i] <- DrugBankData[[i]][[1]]
DrugName[i] <- DrugBankData[[i]][[4]]
}
EdgeListTable = data.frame(DrugName, DB_ID)
write.table(EdgeListTable, file="Output1.txt", quote=F)
The output file looks like this:
X1 X2 X3 X4 X1.1 X2.1 X3.1 X4.1
1 Lepirudin Cetuximab Dornase alfa Denileukin diftitox DB00001 DB00002 DB00003 DB00004
So I am just working on getting this correctly formatted into columns and removing the first line of text from this file, and the "1" at the beginning of the second line...

Answer 1

Thanks for your response, herbaman. I ended up resolving the formatting issues (mostly, except that the columns still aren't aligned...) using the following code:
DrugName_Matrix = matrix(DrugName,nrow=NumNodes,ncol=1)
DrugID_Matrix = matrix(DB_ID,nrow=NumNodes,ncol=1)
Composite_Matrix = cbind(DrugName_Matrix,DrugID_Matrix,Target)
write.table(Composite_Matrix, file="Output1.txt", sep='\t', row.names=F, quote=F)
There remain mysterious column header names ("V1" and "V2") that don't appear in the contents of these two matrices; my attempts to rename them have been unsuccessful using standard methods, e.g.
colnames(Composite_Matrix)[colnames(Composite_Matrix)=="V1"] <- "Drug Name"
colnames(Composite_Matrix)[colnames(Composite_Matrix)=="V2"] <- "Drug ID"
or
setnames(Composite_Matrix, old=c("V1","V2"), new=c("DrugName", "DrugID"))
I'm not sure where these V column headers are originating...
As requested, the contents of the two matrices of interest are:
> DrugName_Matrix
[,1]
[1,] "Lepirudin"
[2,] "Cetuximab"
[3,] "Dornase alfa"
[4,] "Denileukin diftitox"
> DrugID_Matrix
[,1]
[1,] "DB00001"
[2,] "DB00002"
[3,] "DB00003"
[4,] "DB00004"
...and the output table is:
V1 V2
Lepirudin DB00001
Cetuximab DB00002
Dornase alfa DB00003
Denileukin diftitox DB00004

Answer 2

To read drug bank nodes I created the following method:
drug_sub_df <- function(rec, main_node, seconadary_node = NULL, id = "drugbank-id", byValue = FALSE) {
parent_key <- NULL
if (!is.null(id)) {
parent_key <- xmlValue(rec[id][[1]])
}
if (byValue) {
df <- map_df(rec[main_node], xmlValue)
} else {
if (is.null(seconadary_node) && !is.null(rec[[main_node]])) {
df <- xmlToDataFrame(rec[[main_node]], stringsAsFactors = FALSE)
} else {
df <- xmlToDataFrame(rec[[main_node]][[seconadary_node]], stringsAsFactors = FALSE)
}
}
if (nrow(df) > 0 && !is.null(parent_key)) {
df$parent_key <- parent_key
}
return(df)
}
Then I call the method like the following:
# Extract drug enzymes actions df
get_enzymes_actions_df <- function(rec) {
return(map_df(xmlChildren(rec[["enzymes"]]),
~ drug_sub_df(.x, "actions", id = "id")))
}
# Extract drug articles df
get_enzymes_articles_df <- function(rec) {
return(map_df(
xmlChildren(rec[["enzymes"]]),
~ drug_sub_df(.x, "references", seconadary_node = "articles", id = "id")
))
}
Of course. there are different situations that require different solutions like the following:
get_enzyme_rec <- function(r, drug_key) {
tibble(
id = xmlValue(r[["id"]]),
name = xmlValue(r[["name"]]),
organism = xmlValue(r[["organism"]]),
known_action = xmlValue(r[["known-action"]]),
inhibition_strength = xmlValue(r[["inhibition-strength"]]),
induction_strength = xmlValue(r[["induction-strength"]]),
position = ifelse(is.null(xmlGetAttr(r, name = "position")),
NA, xmlGetAttr(r, name = "position")),
parent_key = drug_key
)
}
get_enzymes_df <- function(rec) {
return(map_df(xmlChildren(rec[["enzymes"]]),
~ get_enzyme_rec(.x, xmlValue(rec["drugbank-id"][[1]]))))
}
or that one
get_atc_codes_rec <- function(r, drug_key) {
tibble(
atc_code = xmlGetAttr(r, name = "code"),
level_1 = xmlValue(r[[1]]),
code_1 = xmlGetAttr(r[[1]], name = "code"),
level_2 = xmlValue(r[[2]]),
code_2 = xmlGetAttr(r[[2]], name = "code"),
level_3 = xmlValue(r[[3]]),
code_3 = xmlGetAttr(r[[3]], name = "code"),
level_4 = xmlValue(r[[4]]),
code_4 = xmlGetAttr(r[[4]], name = "code"),
parent_key = drug_key
)
}
get_atc_codes_df <- function(rec) {
return (map_df(xmlChildren(rec[["atc-codes"]]),
~ get_atc_codes_rec(.x,
xmlValue(rec["drugbank-id"][[1]]))))
}
You can find more examples to extract contents of an drug bank XML database in R in different structures in this package
https://github.com/Dainanahan/dbparser