k-mer algorithm

kMer.hs

{-# LANGUAGE UnicodeSyntax #-}

{-
Reads: AGATCGAGTG
Prints: sorted 3-mers:   AGA AGT ATC CGA GAG GAT GTG TCG
        unsorted 3-mers: AGA GAT ATC TCG CGA GAG AGT GTG
-}

import           Data.Function
import           Data.List
import           Data.Ord      ()

kMers ∷ Int → [a] → [[a]]
kMers k seqs = map (take k) $ take (n-k+1) $ tails seqs where n = length seqs

getFreq ∷ Ord a ⇒ [a] → [(a, Int)]
getFreq = sortBy (flip compare `on` snd) . map ((,) <$> head <*> length) . group

topKMers ∷ Ord a ⇒ Int → [a] → Int → [([a], Int)]
topKMers n xs k = take k $ getFreq $ kMers n xs

testSeq ∷ String
testSeq = "AGATCGAGTG"

--main ∷ IO ()
main = do
    let kMerSeq =  kMers 3 testSeq
    let kMerTop =  topKMers 3 testSeq 8

    putStrLn ("k-mers: " ++ unwords kMerSeq)
    print kMerTop