R/find.Overlapping.mzs.R

In BowenNCSU/xMSanalyzer: xMSanalyzer: R package for data analysis, comparison, and annotation of metabolomics data.

#' find.Overlapping.mzs
#' 
#' This function matches features between two datasets using the following user
#' defined criteria: 1) Maximum m/z difference (+/-) ppm 2) Maximum retention
#' time difference in seconds
#' 
#' The find.Overlapping.mzs function operates on two sets of feature lists with
#' m/z and retention times for each feature, denoted by L1 and L2, and iterates
#' over all m/z values in L1 to find those that are within a user defined m/z
#' (ppm) and retention time (sec) threshold in L2. Optionally, the user can
#' match features based on only the m/z values by setting time.thresh=NA. The
#' find.Unique.mzs function uses a similar algorithm to find unique features
#' that are not within a user defined mass and retention time threshold level.
#' 
#' @param dataA apLCMS or XCMS feature table for dataset A
#' @param dataB apLCMS or XCMS feature table for dataset B
#' @param mz.thresh Maximum m/z difference (+/-) ppm. eg: 10
#' @param time.thresh Maximum retention time difference (+/-) secs. eg: 300
#' @param alignment.tool Name of the feature alignment tool eg: "apLCMS" or
#' "XCMS" or "NA" Use "NA" if the input matrix includes only m/z or both m/z
#' and retnetion time values.
#' @return Matrix of overlapping features with columns: index.data.A: index of
#' overlapping m/z in dataset A mz.data.A: m/z in dataset A time.data.A:
#' retention time in dataset A index.data.B: index of overlapping m/z in
#' dataset B mz.data.B: m/z in dataset B time.data.B: retention time in dataset
#' B
#' @author Karan Uppal <kuppal2@@emory.edu>
#' @seealso apLCMS.align, XCMS.align, find.Unique.mzs, getVenn
#' @keywords ~overlap ~common
find.Overlapping.mzs <- function(dataA, dataB, mz.thresh = 10, 
    time.thresh = NA, alignment.tool = NA) {
    
    data_a <- as.data.frame(dataA)
    data_b <- as.data.frame(dataB)
    # data_a<-unique(data_a)
    rm(dataA)
    rm(dataB)
    
    # data_b<-unique(data_b)
    
    com_mz_num = 1
    unique_mz = {
    }
    ppm_v = {
    }
    rt_v = {
    }
    
    commat = {
    }
    
    col.names.dataA = colnames(data_a)
    col.names.dataB = colnames(data_b)
    
    if (is.na(alignment.tool) == FALSE) {
        if (alignment.tool == "apLCMS") {
            sample.col.start = 5
        } else {
            if (alignment.tool == "XCMS") {
                sample.col.start = 9
                col.names.dataA[1] = "mz"
                col.names.dataA[2] = "time"
                col.names.dataB[1] = "mz"
                col.names.dataB[2] = "time"
                colnames(data_a) = col.names.dataA
                colnames(data_b) = col.names.dataB
            }
            
        }
    } else {
        # stop(paste('Invalid value for alignment.tool.
        # Please use either \'apLCMS\' or \'XCMS\'',
        # sep=''))
        
        col.names.dataA[1] = "mz"
        
        col.names.dataB[1] = "mz"
        if (is.na(time.thresh) == FALSE) {
            col.names.dataA[2] = "time"
            col.names.dataB[2] = "time"
            print("Using the 1st columns as \"mz\" and 2nd columsn as \"retention time\"")
        } else {
            print("Using the 1st columns as \"mz\"")
        }
        colnames(data_a) = col.names.dataA
        colnames(data_b) = col.names.dataB
    }
    
    # data_a<-data_a[order(data_a$mz),]
    # data_b<-data_b[order(data_b$mz),]
    data_a <- as.data.frame(data_a)
    data_b <- as.data.frame(data_b)
    colnames(data_a) = col.names.dataA
    colnames(data_b) = col.names.dataB
    # create header for the matrix with common
    # features
    if (is.na(time.thresh) == FALSE) {
        mznames = c("index.A", "mz.data.A", "time.data.A", 
            "index.B", "mz.data.B", "time.data.B", 
            "time.difference")
    } else {
        mznames = c("index.A", "mz.data.A", "index.B", 
            "mz.data.B")
    }
    
    # Step 1 Group features by m/zdim(data_a)[1]
    mz_groups <- lapply(1:dim(data_a)[1], function(j) {
        
        commat = {
        }
        commzA = new("list")
        commzB = new("list")
        ppmb = (mz.thresh) * (data_a$mz[j]/1e+06)
        
        getbind_same <- which(abs(data_b$mz - data_a$mz[j]) <= 
            ppmb)
        
        if (is.na(time.thresh) == FALSE) {
            if (length(getbind_same) > 0) {
                nearest_time_diff = 10000
                bestmatch = {
                }
                rnames = {
                }
                temp = {
                }
                commat = {
                }
                for (comindex in 1:length(getbind_same)) {
                  tempA = cbind(j, data_a[j, c(1, 
                    2)])
                  tempB = cbind(getbind_same[comindex], 
                    data_b[getbind_same[comindex], 
                      c(1, 2)])
                  temp = cbind(tempA, tempB)
                  
                  timediff = abs(data_a[j, 2] - data_b[getbind_same[comindex], 
                    2])
                  
                  temp <- cbind(temp, timediff)
                  
                  if (timediff < time.thresh && timediff <= 
                    nearest_time_diff) {
                    bestmatch = as.data.frame(temp)
                    nearest_time_diff = timediff
                    
                    
                    
                    
                    
                    temp <- as.data.frame(temp)
                    commat <- rbind(commat, temp)
                    rnamestemp <- paste("mz", j, "_", 
                      comindex, sep = "")
                    rnames <- c(rnames, rnamestemp)
                  }
                }
                
                
                # commat=as.data.frame(bestmatch)
                
                if (length(commat) >= 4) {
                  rownames(commat) = rnames
                }
                
                
            }
        } else {
            if (length(getbind_same) > 0) {
                temp1 <- {
                }
                for (comindex in 1:length(getbind_same)) {
                  tempA = cbind(j, data_a[j, c(1)])
                  tempB = cbind(getbind_same[comindex], 
                    data_b[getbind_same[comindex], 
                      c(1)])
                  temp = cbind(tempA, tempB)
                  # temp=cbind(data_a[j,c(1)],data_b[getbind_same[comindex],c(1)])
                  temp1 <- rbind(temp1, temp)
                }
                commat = as.data.frame(temp1)
                rnames <- paste("mz", j, "_", seq(1, 
                  length(getbind_same)), sep = "")
                rownames(commat) = rnames
                
            }
        }
        return(as.data.frame(commat))
        
        
    })
    
    # Step 2 Sub-group features from Step 1 by
    # Retention time find the features with RT values
    # within the defined range as compared to the
    # query feature
    
    uniqueinA = {
    }
    uniqueinB = {
    }
    commat = data.frame()
    
    
    if (length(mz_groups) > 0) {
        for (j in 1:length(mz_groups)) {
            temp_diff = {
            }
            
            if (is.list(mz_groups) == TRUE) {
                tempdata = mz_groups[[j]]
            } else {
                tempdata = mz_groups[j]
            }
            
            if (length(tempdata) > 1) {
                
                colnames(tempdata) = mznames
                tempdata = as.data.frame(t(tempdata))
                
                temp = tempdata
                
                temp = as.data.frame(temp)
                
                if (is.null(commat) == TRUE) {
                  commat = t(temp)
                  
                } else {
                  
                  commat = rbind(commat, t(temp))
                  
                }
                
                
            }
            
            
        }
        
        if (is.null(dim(commat)) == FALSE) {
            commat = as.data.frame(commat)
        }
    }
    
    
    return(commat)
}