CleanMemes.sc

#!/usr/bin/env -S scala-cli shebang

import java.io.{ File, FileInputStream, FileWriter, BufferedWriter }
import scala.io.BufferedSource

/**
*  MEME File Example
*
*  ...
*  MOTIF Hsapiens-jaspar2022-MAX::MYC-MA0059.1
*  letter-probability matrix: alength= 4 w= 11 nsites= 45 E=0.000000
*   0.3333333333  0.0476190476  0.4285714286  0.1904761905
*   0.7142857143  0.0476190476  0.1904761905  0.0476190476
*   0.0952380952  0.4285714286  0.4285714286  0.0476190476
*  
*  MOTIF Hsapiens-jaspar2022-PPARG-MA0066.1
*  letter-probability matrix: alength= 4 w= 20 nsites= 45 E=0.000000
*   0.1071428571  0.2857142857  0.5000000000  0.1071428571
*   0.1071428571  0.0000000000  0.0000000000  0.8928571429
*   0.6785714286  0.0000000000  0.3214285714  0.0000000000
*   0.0000000000  0.0357142857  0.9642857143  0.0000000000
*  ...
*/
val memeFilein = new File("jaspar2022-human.txt")
val memeReader = new BufferedSource(new FileInputStream(memeFilein))

/**
*  Transcription Factor File Example
*
*  ...
*  ZBTB7B
*  MEF2D
*  ETV3
*  USF1
*  ATF6
*  PBX1
*  POU2F1
*  TBX19
*  PRRX1
*  ...
*/
val tFsFilein = new File("tfMicroglia.txt")
val tFsReader = new BufferedSource(new FileInputStream(tFsFilein))

val fileWriter = new FileWriter("writeToMe.txt")
val bufferedWriter = new BufferedWriter(fileWriter)

val memeFileLines = memeReader.getLines.toList
val tFs = tFsReader.getLines.toList

def processMemeList(processed: List[String], unprocessed: List[String], keep: Boolean): List[String] = {
  unprocessed match {
    case Nil => processed
    case x::xs => {
      val newProcessed = processed :+ x
      if x.contains("MOTIF") then {
        val transcriptionFactor = x.split("-")(2)
        if tFs.contains(transcriptionFactor) then {
          processMemeList(newProcessed, xs, true)
        } else {
          processMemeList(processed, xs, false)
        }
      } else {
        keep match {
          case true => processMemeList(newProcessed, xs, true)
          case false => processMemeList(processed, xs, false)
        } 
      }
    }
  }
}

processMemeList(List[String]().empty, memeFileLines, true)
  .foreach{line => bufferedWriter.write(line + "\n")}

bufferedWriter.close()