AMDIS自建库结果导出函数

##########check format of MSL library
checkformat<-function(data){
  colno<-which(data=="")
  colno<-c(2,colno[1:length(colno)-1]+2)
  extractdata<-data[colno]
  for(i in 1:length(extractdata)){
    extractdata[i]<-substr(extractdata[i],1,7)
  }
  check_ex<-which(extractdata!="COMMENT")
  check_data<-colno[check_ex]
  return(check_data)
}
########################
MSLexport<-function(data){
  lib<-data.frame(Compound<-character(0),
                  CAS<-character(0),
                  RT<-numeric(0),
                  RI<-numeric(0),
                  Datafile<-character(0),
                  m1<-numeric(0),
                  m2<-numeric(0),
                  m3<-numeric(0),
                  m4<-numeric(0),
                  m5<-numeric(0))
  
  colnames(lib)<-c("Compound","CAS","RT","RI","Datafile","mZ1","mZ2","mZ3","mz4","mz5")
  colno<-which(data=="")
  ##########
  ##extract first compound information
  Compound<-substr(data[1],6,nchar(data[1]))
  CAS<-substr(data[5],7,nchar(data[5]))
  RT<-substr(data[6],4,nchar(data[6]))
  RI<-substr(data[3],4,nchar(data[3]))
  Datafile<-substr(data[2],34,43)
  #####################
  peaknum<-as.numeric(substr(data[8],13,14))
  mzlist<-data.frame(mz<-rep(0,peaknum),abundance<-rep(0,peaknum))
  colnames(mzlist)<-c("mz","abundance")
  remainder<-peaknum%%5
  shang<-peaknum%/%5
  for(m in 1:shang){
    for(j in 1:5){
      mzlist$mz[j+(m-1)*5]<-as.numeric(substr(data[colno[1]-shang-2+m],(11*j-9),(11*j-7)))
      mzlist$abundance[j+(m-1)*5]<-as.numeric(substr(data[colno[1]-shang-2+m],(11*j-5),(11*j-2)))
    }
  }
  if(remainder!=0)
    for(j in 1:remainder){
      mzlist$mz[5*shang+j]<-as.numeric(substr(data[colno[1]-1],(11*j-9),(11*j-7)))
      mzlist$abundance[5*shang+j]<-as.numeric(substr(data[colno[1]-1],(11*j-5),(11*j-2)))
    }
  mzlist<-mzlist[order(mzlist$abundance,decreasing = T),]
  ###############################
  mz1<-mzlist$mz[1]
  mz2<-mzlist$mz[2]
  mz3<-mzlist$mz[3]
  mz4<-mzlist$mz[4]
  mz5<-mzlist$mz[5]
  ##########
  temp<-cbind(Compound,CAS,RT,RI,Datafile,mz1,mz2,mz3,mz4,mz5)
  
  lib<-rbind(lib,temp)
  #########bulid a loop based on the first compound information
  for(i in 2:length(colno)-1){
    Compound<-substr(data[colno[i]+1],6,nchar(data[colno[i]+1]))
    CAS<-substr(data[colno[i]+5],7,nchar(data[colno[i]+5]))
    RT<-substr(data[colno[i]+6],4,nchar(data[colno[i]+6]))
    RI<-substr(data[colno[i]+3],4,nchar(data[colno[i]+3]))
    Datafile<-substr(data[colno[i]+2],34,43)
    #####################
    peaknum<-as.numeric(substr(data[colno[i]+8],13,14))
    mzlist<-data.frame(mz<-rep(0,peaknum),abundance<-rep(0,peaknum))
    colnames(mzlist)<-c("mz","abundance")
    remainder<-peaknum%%5
    shang<-peaknum%/%5
    for(m in 1:shang){
      for(j in 1:5){
        mzlist$mz[j+(m-1)*5]<-as.numeric(substr(data[colno[i+1]-shang-2+m],(11*j-9),(11*j-7)))
        mzlist$abundance[j+(m-1)*5]<-as.numeric(substr(data[colno[i+1]-shang-2+m],(11*j-5),(11*j-2)))
      }
    }
    if(remainder!=0)
      for(j in 1:remainder){
        mzlist$mz[5*shang+j]<-as.numeric(substr(data[colno[i+1]-1],(11*j-9),(11*j-7)))
        mzlist$abundance[5*shang+j]<-as.numeric(substr(data[colno[i+1]-1],(11*j-5),(11*j-2)))
      }
    mzlist<-mzlist[order(mzlist$abundance,decreasing = T),]
    ###############################
    mz1<-mzlist$mz[1]
    mz2<-mzlist$mz[2]
    mz3<-mzlist$mz[3]
    mz4<-mzlist$mz[4]
    mz5<-mzlist$mz[5]
    ##########
    temp<-cbind(Compound,CAS,RT,RI,Datafile,mz1,mz2,mz3,mz4,mz5)
    lib<-rbind(lib,temp)
    
  }
  lib$RT<-as.numeric(as.character(lib$RT))
  lib$RI<-as.numeric(as.character(lib$RI))
  lib<-lib[order(lib$RT),]
  return(lib)
}
##############example############
source("C:/Users/Administrator/Desktop/Rscript/函数脚本/定性库MSL导出.R")
setwd("C:/Users/Administrator/Desktop")
data<-readLines("15rootstock-free.MSL")
str(data)
check_data<-checkformat(data)
#check format of MSL data according to print hint
lib<-MSLexport(data)
write.table(lib,file="15rootstock-freeMSL.csv",col.names = T,row.names = F,sep=",")
注：若在定性过程中更改过化合物名称，MSL文件中该化合物信息的顺序会乱，需要用记事本打开文件检查格式，因此，使用AMDIS定性时应该尽量不更改名称，用该函数导出结果后如有必要再更改。