module MolSeq where

import Hugs.Observe
import Data.List

data SequenceType = DNA | Protein
	deriving (Show, Eq)
data MolSeq = MolSeq { molSeqType :: SequenceType, molSeqName :: String, molSeqSequenceString :: String }
	deriving (Show, Eq)


string2seq :: String -> String -> MolSeq
string2seq moleculeName seq
		| isProtein seq = (MolSeq Protein moleculeName seq)
    	| otherwise = (MolSeq DNA moleculeName seq)
			where
				isProtein :: String -> Bool
				isProtein [] = False
				isProtein (h:t)
					| (h `elem` "ACGTacgt") = isProtein t
					| otherwise = True


seqName :: MolSeq -> String
seqName mol = (molSeqName mol)

seqType :: MolSeq -> SequenceType
seqName mol = (molSeqType mol)

seqLength :: MolSeq -> Int
seqLength mol = length (molSeqSequenceString mol)

seqDistance :: MolSeq -> MolSeq -> Float
seqDistance mol1 mol2
		| compatibleMolecules mol1 mol2 = calcMolDistance mol1 mol2
		| otherwise = error "'mol1' and 'mol2' are not of the same molecule type"


compatibleMolecules :: MolSeq -> MolSeq -> Bool
compatibleMolecules mol1 mol2 = (molSeqType mol1) = (molSeqType mol2)

calcMolDistance :: MolSeq -> MolSeq -> Float
calcMolDistance mol1 mol2
		| seqType mol1 == DNA = jukesCantor mol1 mol2
		| otherwise = poisson mol1 mol2

jukesCantor :: MolSeq -> MolSeq -> Float
jukesCantor mol1 mol2 = -(3/4)