Bioinformatics.FASTA.hs

module Bioinformatics.FASTA (
  FASTAFormatLine(..),
  buildFASTALines,
  buildFASTALinesWith
) where

import qualified Bioinformatics.DNANucleotide as D
import qualified Data.List.Split as S

data FASTAFormatLine a = FASTAFormatLine {  name :: String,
                                            content :: [a]
                                         } deriving (Show)

stringToFASTAFormat :: String -> FASTAFormatLine D.DNANucleotide
stringToFASTAFormat [] = FASTAFormatLine { name = "", content = [] }
stringToFASTAFormat xs =
  let comps = lines xs
  in FASTAFormatLine { name = head comps, content = map D.charToDNANucleotide . concat $ tail comps }

buildFASTALines :: String -> [FASTAFormatLine D.DNANucleotide]
buildFASTALines = map stringToFASTAFormat . S.splitOn ">"

stringToFASTAFormatWith :: (Char -> a) -> String -> FASTAFormatLine a
stringToFASTAFormatWith _ [] = FASTAFormatLine {name = "", content = [] }
stringToFASTAFormatWith with xs =
  let comps = lines xs
  in FASTAFormatLine { name = head comps, content = map with . concat $ tail comps }

buildFASTALinesWith :: (Char -> a) -> String -> [FASTAFormatLine a]
buildFASTALinesWith with = map (stringToFASTAFormatWith with) . S.splitOn ">"

RNASplicing.hs

import qualified Bioinformatics.DNANucleotide as D
import qualified Bioinformatics.RNANucleotide as R
import qualified Bioinformatics.Utilities as U
import qualified Bioinformatics.Protein as P
import qualified Bioinformatics.FASTA as F

import Data.List.Split (splitOn)

getLines :: String -> IO String
getLines contents = do
  line <- getLine
  if null line
    then return contents
    else getLines $ contents ++ line ++ "\n"

getRNA :: String -> [R.RNANucleotide]
getRNA = map (U.dnaToRNA . D.charToDNANucleotide)

getProtein :: String -> P.Protein
getProtein = P.rnaToAminoAcids . getRNA

removeSubstrings :: String -> [String] -> String
removeSubstrings s xs = foldr (\x a -> concat $ splitOn x a) s xs

generateCodingRegion :: String -> [String] -> P.Protein
generateCodingRegion dna introns = init . getProtein $ removeSubstrings dna introns

main :: IO ()
main = do
  input <- getLines ""
  let fastas = tail $ F.buildFASTALinesWith id input
      contents = map F.content fastas
      dna = head contents
      introns = tail contents
  putStrLn . concatMap show $ generateCodingRegion dna introns