The (incorrect) central dogma of molecular biology (DNA -> RNA -> protein) modeled in F#

Dogma.fs

namespace Dogma

module DNA =

    type Nucleobases =
        | A // Adenine, a Purine
        | C // Cytosine, a Pyrimidine
        | G // Guanine, a Purine
        | T // Thymine, a Pyrimidine

    type Codon = (Nucleobases * Nucleobases * Nucleobases)

    let complement nucleobase =
        match nucleobase with
        | A -> T
        | C -> G
        | G -> C
        | T -> A

    let complementSeq sequence = Seq.map complement sequence

    let parse c =
        match System.Char.ToUpper(c) with
        | 'A' -> Some A
        | 'C' -> Some C
        | 'G' -> Some G
        | 'T' -> Some T
        | _ -> None

    let parseSeq (s: string) =
        s.ToCharArray()
        |> Array.choose parse
        
module RNA =

    type Nucleobases =
        | A // Adenine, a Purine
        | C // Cytosine, a Pyrimidine
        | G // Adenine, a Purine
        | U // Uracil, a Pyrimidine

    type Codon = (Nucleobases * Nucleobases * Nucleobases)

    let complement (nucleobase: Nucleobases) =
        match nucleobase with
        | A -> U
        | C -> G
        | G -> C
        | U -> A

module Transcription =

    let dnaToRna dna =
        match dna with
        | DNA.A -> RNA.U    // RNA has Uracil in place of DNA Thymine
        | DNA.C -> RNA.G
        | DNA.G -> RNA.C
        | DNA.T -> RNA.A

    let dnaSeqToRnaSeq (dna: seq<DNA.Nucleobases>) =
        dna
        |> Seq.map dnaToRna

module Translation = 

    let seqToCodons (s: seq<'T>) =
        let a = s |> Seq.toArray
        seq {
            for i in 0 .. 3 .. a.Length - 3 do
                yield (a.[i], a.[i+1], a.[i+2])
        }
        
    type AminoAcid =
        | Alanine
        | Arginine
        | Asparagine
        | AsparticAcid
        | Cysteine
        | Glutamine
        | GlutamicAcid
        | Glycine
        | Histidine
        | Isoleucine
        | Leucine
        | Lysine
        | Methionine
        | Phenylalanine
        | Proline
        | Serine
        | Threonine
        | Tryptophan
        | Tyrosine
        | Valine

    let aminoAcidToOneLetter =
        function
        | Alanine -> "A"
        | Arginine -> "R"
        | Asparagine -> "N"
        | AsparticAcid -> "D"
        | Cysteine -> "C"
        | Glutamine -> "Q"
        | GlutamicAcid -> "E"
        | Glycine -> "G"
        | Histidine -> "H"
        | Isoleucine -> "I"
        | Leucine -> "L"
        | Lysine -> "K"
        | Methionine -> "M"
        | Phenylalanine -> "F"
        | Proline -> "P"
        | Serine -> "S"
        | Threonine -> "T"
        | Tryptophan -> "W"
        | Tyrosine -> "Y"
        | Valine -> "V"

    type Signal =
        | StartOrAmino of AminoAcid
        | Stop
        | Amino of AminoAcid

    let rnaCodonToSignal =
        function
        // 1st base U
        | (RNA.Nucleobases.U, RNA.Nucleobases.U, RNA.Nucleobases.U) -> Amino Phenylalanine
        | (RNA.Nucleobases.U, RNA.Nucleobases.U, RNA.Nucleobases.C) -> Amino Phenylalanine
        | (RNA.Nucleobases.U, RNA.Nucleobases.U, RNA.Nucleobases.A) -> Amino Leucine
        | (RNA.Nucleobases.U, RNA.Nucleobases.U, RNA.Nucleobases.G) -> StartOrAmino Leucine

        | (RNA.Nucleobases.U, RNA.Nucleobases.C, RNA.Nucleobases.U) -> Amino Serine
        | (RNA.Nucleobases.U, RNA.Nucleobases.C, RNA.Nucleobases.C) -> Amino Serine
        | (RNA.Nucleobases.U, RNA.Nucleobases.C, RNA.Nucleobases.A) -> Amino Serine
        | (RNA.Nucleobases.U, RNA.Nucleobases.C, RNA.Nucleobases.G) -> Amino Serine

        | (RNA.Nucleobases.U, RNA.Nucleobases.A, RNA.Nucleobases.U) -> Amino Tyrosine
        | (RNA.Nucleobases.U, RNA.Nucleobases.A, RNA.Nucleobases.C) -> Amino Tyrosine
        | (RNA.Nucleobases.U, RNA.Nucleobases.A, RNA.Nucleobases.A) -> Stop
        | (RNA.Nucleobases.U, RNA.Nucleobases.A, RNA.Nucleobases.G) -> Stop

        | (RNA.Nucleobases.U, RNA.Nucleobases.G, RNA.Nucleobases.U) -> Amino Cysteine
        | (RNA.Nucleobases.U, RNA.Nucleobases.G, RNA.Nucleobases.C) -> Amino Cysteine
        | (RNA.Nucleobases.U, RNA.Nucleobases.G, RNA.Nucleobases.A) -> Stop
        | (RNA.Nucleobases.U, RNA.Nucleobases.G, RNA.Nucleobases.G) -> Amino Tryptophan

        // 1st base C
        | (RNA.Nucleobases.C, RNA.Nucleobases.U, RNA.Nucleobases.U) -> Amino Leucine
        | (RNA.Nucleobases.C, RNA.Nucleobases.U, RNA.Nucleobases.C) -> Amino Leucine
        | (RNA.Nucleobases.C, RNA.Nucleobases.U, RNA.Nucleobases.A) -> Amino Leucine
        | (RNA.Nucleobases.C, RNA.Nucleobases.U, RNA.Nucleobases.G) -> Amino Leucine

        | (RNA.Nucleobases.C, RNA.Nucleobases.C, RNA.Nucleobases.U) -> Amino Proline
        | (RNA.Nucleobases.C, RNA.Nucleobases.C, RNA.Nucleobases.C) -> Amino Proline
        | (RNA.Nucleobases.C, RNA.Nucleobases.C, RNA.Nucleobases.A) -> Amino Proline
        | (RNA.Nucleobases.C, RNA.Nucleobases.C, RNA.Nucleobases.G) -> Amino Proline

        | (RNA.Nucleobases.C, RNA.Nucleobases.A, RNA.Nucleobases.U) -> Amino Histidine
        | (RNA.Nucleobases.C, RNA.Nucleobases.A, RNA.Nucleobases.C) -> Amino Histidine
        | (RNA.Nucleobases.C, RNA.Nucleobases.A, RNA.Nucleobases.A) -> Amino Glutamine
        | (RNA.Nucleobases.C, RNA.Nucleobases.A, RNA.Nucleobases.G) -> Amino Glutamine

        | (RNA.Nucleobases.C, RNA.Nucleobases.G, RNA.Nucleobases.U) -> Amino Arginine
        | (RNA.Nucleobases.C, RNA.Nucleobases.G, RNA.Nucleobases.C) -> Amino Arginine
        | (RNA.Nucleobases.C, RNA.Nucleobases.G, RNA.Nucleobases.A) -> Amino Arginine
        | (RNA.Nucleobases.C, RNA.Nucleobases.G, RNA.Nucleobases.G) -> Amino Arginine

        // 1st base A
        | (RNA.Nucleobases.A, RNA.Nucleobases.U, RNA.Nucleobases.U) -> Amino Isoleucine
        | (RNA.Nucleobases.A, RNA.Nucleobases.U, RNA.Nucleobases.C) -> Amino Isoleucine
        | (RNA.Nucleobases.A, RNA.Nucleobases.U, RNA.Nucleobases.A) -> Amino Isoleucine
        | (RNA.Nucleobases.A, RNA.Nucleobases.U, RNA.Nucleobases.G) -> StartOrAmino Methionine

        | (RNA.Nucleobases.A, RNA.Nucleobases.C, RNA.Nucleobases.U) -> Amino Threonine
        | (RNA.Nucleobases.A, RNA.Nucleobases.C, RNA.Nucleobases.C) -> Amino Threonine
        | (RNA.Nucleobases.A, RNA.Nucleobases.C, RNA.Nucleobases.A) -> Amino Threonine
        | (RNA.Nucleobases.A, RNA.Nucleobases.C, RNA.Nucleobases.G) -> Amino Threonine

        | (RNA.Nucleobases.A, RNA.Nucleobases.A, RNA.Nucleobases.U) -> Amino Asparagine
        | (RNA.Nucleobases.A, RNA.Nucleobases.A, RNA.Nucleobases.C) -> Amino Asparagine
        | (RNA.Nucleobases.A, RNA.Nucleobases.A, RNA.Nucleobases.A) -> Amino Lysine
        | (RNA.Nucleobases.A, RNA.Nucleobases.A, RNA.Nucleobases.G) -> Amino Lysine

        | (RNA.Nucleobases.A, RNA.Nucleobases.G, RNA.Nucleobases.U) -> Amino Serine
        | (RNA.Nucleobases.A, RNA.Nucleobases.G, RNA.Nucleobases.C) -> Amino Serine
        | (RNA.Nucleobases.A, RNA.Nucleobases.G, RNA.Nucleobases.A) -> Amino Arginine
        | (RNA.Nucleobases.A, RNA.Nucleobases.G, RNA.Nucleobases.G) -> Amino Arginine

        // 1st base G
        | (RNA.Nucleobases.G, RNA.Nucleobases.U, RNA.Nucleobases.U) -> Amino Valine
        | (RNA.Nucleobases.G, RNA.Nucleobases.U, RNA.Nucleobases.C) -> Amino Valine
        | (RNA.Nucleobases.G, RNA.Nucleobases.U, RNA.Nucleobases.A) -> Amino Valine
        | (RNA.Nucleobases.G, RNA.Nucleobases.U, RNA.Nucleobases.G) -> StartOrAmino Valine

        | (RNA.Nucleobases.G, RNA.Nucleobases.C, RNA.Nucleobases.U) -> Amino Alanine
        | (RNA.Nucleobases.G, RNA.Nucleobases.C, RNA.Nucleobases.C) -> Amino Alanine
        | (RNA.Nucleobases.G, RNA.Nucleobases.C, RNA.Nucleobases.A) -> Amino Alanine
        | (RNA.Nucleobases.G, RNA.Nucleobases.C, RNA.Nucleobases.G) -> Amino Alanine

        | (RNA.Nucleobases.G, RNA.Nucleobases.A, RNA.Nucleobases.U) -> Amino AsparticAcid
        | (RNA.Nucleobases.G, RNA.Nucleobases.A, RNA.Nucleobases.C) -> Amino AsparticAcid
        | (RNA.Nucleobases.G, RNA.Nucleobases.A, RNA.Nucleobases.A) -> Amino GlutamicAcid
        | (RNA.Nucleobases.G, RNA.Nucleobases.A, RNA.Nucleobases.G) -> Amino GlutamicAcid

        | (RNA.Nucleobases.G, RNA.Nucleobases.G, RNA.Nucleobases.U) -> Amino Glycine
        | (RNA.Nucleobases.G, RNA.Nucleobases.G, RNA.Nucleobases.C) -> Amino Glycine
        | (RNA.Nucleobases.G, RNA.Nucleobases.G, RNA.Nucleobases.A) -> Amino Glycine
        | (RNA.Nucleobases.G, RNA.Nucleobases.G, RNA.Nucleobases.G) -> Amino Glycine

    let dnaCodonToSignal =
        function
        // 1st base T
        | (DNA.Nucleobases.T, DNA.Nucleobases.T, DNA.Nucleobases.T) -> Amino Phenylalanine
        | (DNA.Nucleobases.T, DNA.Nucleobases.T, DNA.Nucleobases.C) -> Amino Phenylalanine
        | (DNA.Nucleobases.T, DNA.Nucleobases.T, DNA.Nucleobases.A) -> Amino Leucine
        | (DNA.Nucleobases.T, DNA.Nucleobases.T, DNA.Nucleobases.G) -> StartOrAmino Leucine

        | (DNA.Nucleobases.T, DNA.Nucleobases.C, DNA.Nucleobases.T) -> Amino Serine
        | (DNA.Nucleobases.T, DNA.Nucleobases.C, DNA.Nucleobases.C) -> Amino Serine
        | (DNA.Nucleobases.T, DNA.Nucleobases.C, DNA.Nucleobases.A) -> Amino Serine
        | (DNA.Nucleobases.T, DNA.Nucleobases.C, DNA.Nucleobases.G) -> Amino Serine

        | (DNA.Nucleobases.T, DNA.Nucleobases.A, DNA.Nucleobases.T) -> Amino Tyrosine
        | (DNA.Nucleobases.T, DNA.Nucleobases.A, DNA.Nucleobases.C) -> Amino Tyrosine
        | (DNA.Nucleobases.T, DNA.Nucleobases.A, DNA.Nucleobases.A) -> Stop
        | (DNA.Nucleobases.T, DNA.Nucleobases.A, DNA.Nucleobases.G) -> Stop

        | (DNA.Nucleobases.T, DNA.Nucleobases.G, DNA.Nucleobases.T) -> Amino Cysteine
        | (DNA.Nucleobases.T, DNA.Nucleobases.G, DNA.Nucleobases.C) -> Amino Cysteine
        | (DNA.Nucleobases.T, DNA.Nucleobases.G, DNA.Nucleobases.A) -> Stop
        | (DNA.Nucleobases.T, DNA.Nucleobases.G, DNA.Nucleobases.G) -> Amino Tryptophan

        // 1st base C
        | (DNA.Nucleobases.C, DNA.Nucleobases.T, DNA.Nucleobases.T) -> Amino Leucine
        | (DNA.Nucleobases.C, DNA.Nucleobases.T, DNA.Nucleobases.C) -> Amino Leucine
        | (DNA.Nucleobases.C, DNA.Nucleobases.T, DNA.Nucleobases.A) -> Amino Leucine
        | (DNA.Nucleobases.C, DNA.Nucleobases.T, DNA.Nucleobases.G) -> Amino Leucine

        | (DNA.Nucleobases.C, DNA.Nucleobases.C, DNA.Nucleobases.T) -> Amino Proline
        | (DNA.Nucleobases.C, DNA.Nucleobases.C, DNA.Nucleobases.C) -> Amino Proline
        | (DNA.Nucleobases.C, DNA.Nucleobases.C, DNA.Nucleobases.A) -> Amino Proline
        | (DNA.Nucleobases.C, DNA.Nucleobases.C, DNA.Nucleobases.G) -> Amino Proline

        | (DNA.Nucleobases.C, DNA.Nucleobases.A, DNA.Nucleobases.T) -> Amino Histidine
        | (DNA.Nucleobases.C, DNA.Nucleobases.A, DNA.Nucleobases.C) -> Amino Histidine
        | (DNA.Nucleobases.C, DNA.Nucleobases.A, DNA.Nucleobases.A) -> Amino Glutamine
        | (DNA.Nucleobases.C, DNA.Nucleobases.A, DNA.Nucleobases.G) -> Amino Glutamine

        | (DNA.Nucleobases.C, DNA.Nucleobases.G, DNA.Nucleobases.T) -> Amino Arginine
        | (DNA.Nucleobases.C, DNA.Nucleobases.G, DNA.Nucleobases.C) -> Amino Arginine
        | (DNA.Nucleobases.C, DNA.Nucleobases.G, DNA.Nucleobases.A) -> Amino Arginine
        | (DNA.Nucleobases.C, DNA.Nucleobases.G, DNA.Nucleobases.G) -> Amino Arginine

        // 1st base A
        | (DNA.Nucleobases.A, DNA.Nucleobases.T, DNA.Nucleobases.T) -> Amino Isoleucine
        | (DNA.Nucleobases.A, DNA.Nucleobases.T, DNA.Nucleobases.C) -> Amino Isoleucine
        | (DNA.Nucleobases.A, DNA.Nucleobases.T, DNA.Nucleobases.A) -> Amino Isoleucine
        | (DNA.Nucleobases.A, DNA.Nucleobases.T, DNA.Nucleobases.G) -> StartOrAmino Methionine

        | (DNA.Nucleobases.A, DNA.Nucleobases.C, DNA.Nucleobases.T) -> Amino Threonine
        | (DNA.Nucleobases.A, DNA.Nucleobases.C, DNA.Nucleobases.C) -> Amino Threonine
        | (DNA.Nucleobases.A, DNA.Nucleobases.C, DNA.Nucleobases.A) -> Amino Threonine
        | (DNA.Nucleobases.A, DNA.Nucleobases.C, DNA.Nucleobases.G) -> Amino Threonine

        | (DNA.Nucleobases.A, DNA.Nucleobases.A, DNA.Nucleobases.T) -> Amino Asparagine
        | (DNA.Nucleobases.A, DNA.Nucleobases.A, DNA.Nucleobases.C) -> Amino Asparagine
        | (DNA.Nucleobases.A, DNA.Nucleobases.A, DNA.Nucleobases.A) -> Amino Lysine
        | (DNA.Nucleobases.A, DNA.Nucleobases.A, DNA.Nucleobases.G) -> Amino Lysine

        | (DNA.Nucleobases.A, DNA.Nucleobases.G, DNA.Nucleobases.T) -> Amino Serine
        | (DNA.Nucleobases.A, DNA.Nucleobases.G, DNA.Nucleobases.C) -> Amino Serine
        | (DNA.Nucleobases.A, DNA.Nucleobases.G, DNA.Nucleobases.A) -> Amino Arginine
        | (DNA.Nucleobases.A, DNA.Nucleobases.G, DNA.Nucleobases.G) -> Amino Arginine

        // 1st base G
        | (DNA.Nucleobases.G, DNA.Nucleobases.T, DNA.Nucleobases.T) -> Amino Valine
        | (DNA.Nucleobases.G, DNA.Nucleobases.T, DNA.Nucleobases.C) -> Amino Valine
        | (DNA.Nucleobases.G, DNA.Nucleobases.T, DNA.Nucleobases.A) -> Amino Valine
        | (DNA.Nucleobases.G, DNA.Nucleobases.T, DNA.Nucleobases.G) -> StartOrAmino Valine

        | (DNA.Nucleobases.G, DNA.Nucleobases.C, DNA.Nucleobases.T) -> Amino Alanine
        | (DNA.Nucleobases.G, DNA.Nucleobases.C, DNA.Nucleobases.C) -> Amino Alanine
        | (DNA.Nucleobases.G, DNA.Nucleobases.C, DNA.Nucleobases.A) -> Amino Alanine
        | (DNA.Nucleobases.G, DNA.Nucleobases.C, DNA.Nucleobases.G) -> Amino Alanine

        | (DNA.Nucleobases.G, DNA.Nucleobases.A, DNA.Nucleobases.T) -> Amino AsparticAcid
        | (DNA.Nucleobases.G, DNA.Nucleobases.A, DNA.Nucleobases.C) -> Amino AsparticAcid
        | (DNA.Nucleobases.G, DNA.Nucleobases.A, DNA.Nucleobases.A) -> Amino GlutamicAcid
        | (DNA.Nucleobases.G, DNA.Nucleobases.A, DNA.Nucleobases.G) -> Amino GlutamicAcid

        | (DNA.Nucleobases.G, DNA.Nucleobases.G, DNA.Nucleobases.T) -> Amino Glycine
        | (DNA.Nucleobases.G, DNA.Nucleobases.G, DNA.Nucleobases.C) -> Amino Glycine
        | (DNA.Nucleobases.G, DNA.Nucleobases.G, DNA.Nucleobases.A) -> Amino Glycine
        | (DNA.Nucleobases.G, DNA.Nucleobases.G, DNA.Nucleobases.G) -> Amino Glycine

    let dnaCodonsToSignals codons =
        codons
        |> Seq.map dnaCodonToSignal

    let rnaCodonsToSignals codons =
        codons
        |> Seq.map rnaCodonToSignal

    let rnaSeqToSignals rna =
        rna
        |> seqToCodons
        |> rnaCodonsToSignals

module Dogma =

    let dogma dnaSeq =
        dnaSeq
        |> Transcription.dnaSeqToRnaSeq
        |> Translation.rnaSeqToSignals

module PrettyPrint =

    open System.Text
    open Translation

    let signalsToString (signals: seq<Translation.Signal>) =
        signals
        |> Seq.map (fun s -> match s with
                             | StartOrAmino a -> aminoAcidToOneLetter a
                             | Amino a -> aminoAcidToOneLetter a
                             | Stop -> "-STOP-")
        |> Seq.fold (fun (b: StringBuilder) s -> b.Append(s)) (StringBuilder())
        |> fun s -> s.ToString()