Import DrugBank to SQLite

DrugBank2SQLite.R

########################################
## Import of DrugBank Annotation Data ##
########################################
## Function to import DrugBank XML to data.frame
## Last step gives error. To debug, the following function may help.
## Note, this functions needs some major speed improvements. Ideally,
## it should be replaced with a standard XML import method. 
## (1) Download
##     - download DrugBank XML (https://www.drugbank.ca/releases/latest)
##     - name uncompressed file 'drugbank.xml'
## (2) Function to convert XML to data.frame
##     - for debugging the function does it for each entry in loop
dbxml2df <- function(xmlfile) {
    require(XML)
    myxml <- xmlParse(file = xmlfile)
    rootnode <- xmlRoot(myxml)
    rootsize <- xmlSize(rootnode)
    ## (a) Identify which columns are necessary by extracting them for each entry in loop
    cat("\n", "Extracting column names. This may take some time.", "\n")
    mycol <- NULL
    for(i in 1:rootsize) mycol <- unique(c(mycol, colnames(xmlToDataFrame(rootnode[i]))))
    mycol <- as.character(na.omit(mycol))
    cat("\n", "The following columns were identified:", "\n", mycol, "\n")

    ## (b) Extract corresponding data in loop and inject into preformatted data.frame 
    cat("\n", "Extracting data for column names. This may take some time.", "\n")
    df <- as.data.frame(matrix(NA, nrow=rootsize, ncol=length(mycol), dimnames=list(1:rootsize, mycol)))
    for(i in 1:rootsize) {
        tmp <- xmlToDataFrame(rootnode[i], stringsAsFactors = FALSE)
        tmp <- tmp[, !is.na(colnames(tmp))] # removes columns named NA
        v <- as.character(tmp[1,]); names(v) <- colnames(tmp)
        df[i,] <- v[mycol]
    }
    return(df)
}

## Usage:
dbdf <- dbxml2df(xmlfile="data/drugbank.xml")