R/make.known.table.R

In yufree/apLCMS: Adaptive processing of LC-MS data

make.known.table <-
function(metabolite.table, adduct.table, ion.mode="+")
###		the metabolite.table should be a four-column data frame. The columns should be chemical formula, HMDB_ID, KEGG ID, and monoisotopic mass
###		The two IDs can be substitute by others
###		the adduct table should be a four-column data frame. The columns are adduct type, mass divider (absolute value of the number of charges), mass addition (after dividing), and charge of the ion form.
{
    
    metabolite.table<-metabolite.table[order(metabolite.table[,4]),]
    for(i in 1:3) metabolite.table[,i]<-as.vector(metabolite.table[,i])
    
    metabolite.table<-cbind(metabolite.table, matrix(NA, nrow=nrow(metabolite.table),ncol=14))
    colnames(metabolite.table)[5:ncol(metabolite.table)]<-c("ion.type","m.z","Number_profiles_processed","Percent_found","mz_min","mz_max","RT_mean","RT_sd","RT_min","RT_max","int_mean(log)","int_sd(log)","int_min(log)","int_max(log)")
    
    ###		step 1:take out charged ions
    
    l<-nchar(metabolite.table[,1])
    last.char<-substr(metabolite.table[,1], l, l)
    sel<-which(last.char=="+" | last.char=="-")
    
    rm.first.row<-TRUE
    new.table<-matrix(NA, nrow=1, ncol=ncol(metabolite.table))
    colnames(new.table)<-colnames(metabolite.table)
    
    if(length(sel)>0)
    {
        new.table<-rbind(new.table, metabolite.table[sel,])
        
        l<-nchar(new.table[,1])-1
        num.charge<-substr(new.table[,1], l, l)
        num.charge<-as.numeric(num.charge)
        
        new.table[,5]<-"orig"
        new.table[,6]<-new.table[,4]
        new.table[which(!is.na(num.charge)),6]<-new.table[which(!is.na(num.charge)),6]/num.charge[which(!is.na(num.charge))]
        
        metabolite.table<-metabolite.table[-sel,]
    }
    
    
    ###		step 2: merge metaoblites with same m/z values into a single row
    
    n<-1
    m<-2
    to.remove<-rep(0, nrow(metabolite.table))
    
    while(m <= nrow(metabolite.table))
    {
        if(abs(metabolite.table[m,4]-metabolite.table[n,4])<1e-10)
        {
            if(metabolite.table[n,1] != metabolite.table[m,1]) metabolite.table[n,1]<-paste(metabolite.table[n,1],metabolite.table[m,1], sep="/")
            for(j in 2:3) metabolite.table[n,j]<-paste(metabolite.table[n,j],metabolite.table[m,j], sep="/")
            to.remove[m]<-1
            m<-m+1
        }else{
            n<-m
            m<-m+1
        }
        cat("(", n, m, ")")
    }
    
    if(sum(to.remove)>0) metabolite.table<-metabolite.table[-which(to.remove==1),]
    
    
    ###		generate ion adducts for uncharged metabolites, merge results to the new table
    
    neutral.table<-metabolite.table
    
    for(n in 1:nrow(adduct.table))
    {
        this<-neutral.table
        this[,5]<-adduct.table[n,1]
        this[,6]<-neutral.table[,4]/adduct.table[n,2]+adduct.table[n,3]
        
        new.table<-rbind(new.table, this)
    }
    if(rm.first.row) new.table<-new.table[-1,]
    
    ###		ion mode
    
    to.remove<-rep(0, nrow(new.table))
    l<-nchar(new.table[,1])
    last.char<-substr(new.table[,1], l, l)
    if(ion.mode == "+") to.remove[which(last.char == "-")]<-1
    if(ion.mode == "-") to.remove[which(last.char == "+")]<-1
    if(sum(to.remove)>0) new.table<-new.table[-which(to.remove==1),]
    
    ###		are there m/z overlaps?
    
    new.table<-new.table[order(new.table[,6]),]
    
    n<-1
    m<-2
    to.remove<-rep(0, nrow(new.table))
    
    while(m <= nrow(new.table))
    {
        if(abs(new.table[m,6]-new.table[n,6])<1e-10)
        {
            if(new.table[n,1] != new.table[m,1]) new.table[n,1]<-paste(new.table[n,1],new.table[m,1], sep="/")
            for(j in 2:5) new.table[n,j]<-paste(new.table[n,j],new.table[m,j], sep="/")
            to.remove[m]<-1
            m<-m+1
        }else{
            n<-m
            m<-m+1
        }
        cat("*(", n, m, ")")
    }
    
    if(sum(to.remove)>0) new.table<-new.table[-which(to.remove==1),]
    
    ###		output the table
    return(new.table)
}