2_HW_BioInf

import random

def makeDNA(countAT, countCG):
    countA = int(countAT / 2)
    countT = int(countAT / 2)
    countC = int(countCG / 2)
    countG = int(countCG / 2)
    strA = ''.join(random.choice("A") for _ in range(countA))
    strT = ''.join(random.choice("T") for _ in range(countT))
    strC = ''.join(random.choice("C") for _ in range(countC))
    strG = ''.join(random.choice("G") for _ in range(countG))
    listDNA = list(strA + strT + strC + strG)
    random.shuffle(listDNA)
    DNA = ''.join(listDNA)
    return DNA


def findORF(strDNA):
    listORF = list(strDNA[0 + i:3 + i] for i in range(0, len(strDNA), 3))

    for i in range(len(listORF)):
        newORF = []
        if listORF[i] == "ATG":
            newORF.append(listORF[i])
            for j in range(i + 1, len(listORF)):
                if listORF[j] != 'TAA' and listORF[j] != 'TAG' and listORF[j] != 'TGA' and listORF[j] != 'ATG':
                    newORF.append(listORF[j])
                else:
                    if listORF[j] != 'ATG':
                        newORF.append(listORF[j])
                        return newORF


def secondFrame(strDNA):
    strDNA2 = strDNA[1:]
    return findORF(strDNA2)


def thirdFrame(strDNA):
    strDNA2 = strDNA[2:]
    return findORF(strDNA2)


def check(newORF):
    if newORF is None:
        return 0
    else:
        strORF = ''
        for i in range(0, len(newORF)):
            strORF = strORF + newORF[i]

        if len(strORF) % 3 == 0 and len(strORF) >= 90:
            return 1
        else:
            return 0


def reverseDNA(strDNA):
    rDNA = ""
    for i in strDNA:
        if i == "A":
            rDNA = rDNA + "G"
        elif i == "T":
            rDNA = rDNA + "C"
        if i == "G":
            rDNA = rDNA + "A"
        elif i == "C":
            rDNA = rDNA + "T"
    revDNA = rDNA[::-1]
    return revDNA


def print3(frame):
    newStr = ''
    for i in frame:
        newStr += i + ' '
    return newStr


if __name__ == '__main__':
    dnaSize = 1000
    for i in range(20, 81):
        k = 0
        for j in range(1, 10000):
            percentGC = i
            percentAT = 100 - percentGC
            countGC = int((dnaSize * percentGC) / 100)
            countAT = int((dnaSize * percentAT) / 100)
            strDNA = makeDNA(countAT, countGC)
            dfirstFrame = findORF(strDNA)
            dsecondFrame = secondFrame(strDNA)
            dthirdFrame = thirdFrame(strDNA)
            rDNA = reverseDNA(strDNA)
            rfirstFrame = findORF(rDNA)
            rsecondFrame = secondFrame(rDNA)
            rthirdFrame = thirdFrame(rDNA)
            if check(dfirstFrame) | check(dsecondFrame) | check(dthirdFrame) | check(rfirstFrame) | check(rfirstFrame) | check(rsecondFrame) | check(rthirdFrame):
                k += 1

        print(percentGC,";", k / 10000)