HMP转Flapjack基因型和图谱文件

【功能】

将HMP文件转换为Flapjack的genotype和map文件。

【测试】

Null

【更新历史】

2020年2月23日

• 自动完成转换，并在输入文件相同目录下生成【flapjackGeno.txt】文件和【flapjackmap.txt】。

【使用方法】

在Rstudio中运行下列代码，选择要转换的hmp文件。

# Hmp converted to flapjack genotype file for import
rm(list=ls())

### function section ###
# the function of Read files
readFiles <- function(header=TRUE,choose=FALSE,fname){
  if(!"readr" %in% installed.packages()) {
    install.packages("readr")
    library(readr)
  }else{
    library(readr)
  }
  if (choose==TRUE){
    myFileName <- file.choose()
  }else{
    myFileName <- fname
  }
  if (grepl("\\.csv$",myFileName)){
    if (header==TRUE){
      myFile <- read_csv(myFileName, col_names = TRUE)
    }else{
      myFile <- read_csv(myFileName, col_names = FALSE)
    }
  }else{
    if (header==TRUE){
      myFile <- read_tsv(myFileName, col_names = TRUE)
    }else{
      myFile <- read_tsv(myFileName, col_names = FALSE)
    }
  }
  return(myFile)
}

# the function of replace the letters of hybrids
repHybrids <- function(myVector){
  myVector <- sub("R","A/G",myVector)
  myVector <- sub("Y","C/T",myVector)
  myVector <- sub("S","C/G",myVector)
  myVector <- sub("W","A/T",myVector)
  myVector <- sub("K","G/T",myVector)
  myVector <- sub("M","A/C",myVector)
  myVector <- sub("N","-",myVector)
  return(myVector)
}
### function section end ###

### read file ###  
# choose a file of HMP
fn <- choose.files()

# set workspace
setwd(dirname(fn))

# read a file of HMP
hmpFile <- readFiles(header=T,fname(basename(fn)))

### read file end ###

### genotype ###
# remove useless columns
myHmpFile <- as.matrix(hmpFile[,c(1,12:ncol(hmpFile))])

# replace the hybrids
system.time(newHmpFile <- apply(myHmpFile,2,repHybrids))

# check consistancy
all(myHmpFile[,1] == newHmpFile[,1])
all(colnames(myHmpFile) == colnames(newHmpFile))

# Transposing
resultImprotFile <- t(newHmpFile)
resultImprotFile <- cbind(rownames(resultImprotFile),resultImprotFile)
resultImprotFile[,1] <- sub("rs#","",rownames(resultImprotFile))
resultImprotFile <- rbind("",resultImprotFile)
resultImprotFile[1,1] <- "# fjFile = GENOTYPE"

# output file
write.table(resultImprotFile,"flapjackGeno.txt",quote = FALSE,sep="\t",row.names=FALSE,col.names = FALSE)
### genotype end ###

### map ###
# map file
myMapFile <- as.matrix(hmpFile[,c(1,3,4)])
myMapFile[,3] <- as.numeric(myMapFile[,3])
myMapFile <- rbind("",myMapFile)
myMapFile[1,1] <- "# fjFile = MAP"
# output file
write.table(myMapFile,"flapjackMap.txt",quote = FALSE,sep="\t",row.names=FALSE,col.names = FALSE)
### map end ###