main.cpp

#include <iostream>
#include <cassert>
#include <string>
#include <cmath>
using namespace std;

bool isValidBase(char base){
    if (base == 'A' || base == 'C' || base == 'G' || base == 'T'){
        return true;
    }else{
        return false;
    }
}


bool isValidStrand(string strand){
    if (strand == ""){
        return false;
    }

    for (char n : strand){
        if (isValidBase(n) == false){
            return false;
        }
    }
    return true;
}


double strandSimilarity(string strand1, string strand2){
    if (strand1.length() != strand2.length()){
        return 0;
    }else if(strand1 == "" || strand2 == ""){
        return 0;
    }

// why use size_t?

    size_t i = 0; 
    double similar = 0.0, count = 0.0;
    while (i < strand1.length()){
        if (strand1[i] == strand2[i]){
            similar ++;
        }
        count ++;
        i ++;
    }
    return similar / count;
}


int bestStrandMatch(string input_strand, string target_strand){
    //define similarity score:  
    double similarity = 0;
    int position = 0;

    //check if input lenght is longer than target length
    if (input_strand.length() < target_strand.length()) {
        cout << "Best similarity score: " <<  "0.0" << endl;
        return -1;
    }
    
    //
    for (unsigned int i = 0; i <= input_strand.length() - target_strand.length(); i ++){
        double score = strandSimilarity(input_strand.substr(i, target_strand.length()), target_strand);
    
    if (score > similarity){
        similarity = score;
        position = i;
    }
    }

    cout << "Best similarity score: " << similarity << endl;
    return position;
}


void identifyMutations(string input_strand, string target_strand){

// define variables:
    string substring1, substring2;
    double similarity = strandSimilarity(substring1, substring2);
    int alignment;

// define alignment while also prints similarity for most cases

    if (input_strand.size() > target_strand.size()){
        alignment = bestStrandMatch(input_strand, target_strand);
    }else if(input_strand.size() < target_strand.size()){
        alignment = bestStrandMatch(target_strand, input_strand);
    }else if(input_strand == target_strand){ // for completely similar cases
        similarity = 1;
        cout << "Best similarity score: " << similarity << endl;
        alignment = 0;
    }else{ //for cases that only need substitution
        alignment = bestStrandMatch(target_strand, input_strand);
    }

// define substring1 and substring 2
// first find substring so the two strings are the same length. 
    if (input_strand.size() > target_strand.size()){
        substring1 = input_strand.substr(alignment, target_strand.length());
        substring2 = target_strand;
    }else if(input_strand.size() < target_strand.size()){
        substring1 = input_strand;
        substring2 = target_strand.substr(alignment, input_strand.length());
    }else{
        substring1 = input_strand;
        substring2 = target_strand;
    }

// print alignment:
    cout << "Best alignment index: " << alignment << endl;


//check for insertion and deletion BEFORE substring1 and substring 2
    if (input_strand.size() < target_strand.size()){
        for (int i = 0; i < alignment; i++){
            cout << "Insertion at position "<< i + 1<< ": " << target_strand[i] << " is inserted in target strand" << endl;
        }
    }else if(input_strand.size() > target_strand.size()){
        for (int i = 0; i < alignment; i++){
            cout << "Deletion at position "<< i + 1  << ": " << input_strand[i] << " is deleted in target strand" << endl;
        }
    }

//substitution in the MIDDLE
    for (unsigned int i = 0; i < substring1.length(); i ++){
        if (substring1[i] != substring2[i]){
            cout << "Substitution at position " << alignment + i + 1 << ": " << substring1[i] << " -> " << substring2[i] << endl;
        }
    }

//check for insertion and deletion AFTER substring1 and substring 2
    if (input_strand.size() < target_strand.size()){
        for (unsigned int i = alignment + input_strand.length(); i < target_strand.length(); i++){
            cout << "Insertion at position "<< i + 1<< ": " << target_strand[i] << " is inserted in target strand" << endl;
        }
    }else if(input_strand.size() > target_strand.size()){
        for (unsigned int i = alignment + target_strand.length(); i < input_strand.length(); i++){
            cout << "Deletion at position "<< i + 1  << ": " << input_strand[i] << " is deleted in target strand" << endl;
        }
    }else if(similarity == 1){
        cout << "No mutations found." << endl;
        return;
    }

}


void transcribeDNAtoRNA(string strand){

    for (unsigned int i = 0; i < strand.length(); i++){
        if (strand[i] == 'T'){
            cout << 'U';
        }else{
            cout << strand[i];
        }
    }
}


void reverseComplement(string strand){

    for (int i = strand.length(); i >= 0; i--){
        if (strand[i] == 'A'){
            cout << 'T';
        }else if (strand[i] == 'T'){
            cout << 'A';
        }else if (strand[i] == 'C'){
            cout << 'G';
        }else if (strand[i] == 'G'){
            cout << 'C';
        }
    }
}


void getCodingFrames(string strand) {
    bool found = false;

    for (unsigned int i = 0; i <= strand.length() - 3; i++) {
        if (strand.substr(i, 3) == "ATG") {  // Start codon found
            unsigned int j = i + 3;

            while (j <= strand.length() - 3) {
                string codon = strand.substr(j, 3);
                
                if (codon == "TAA" || codon == "TAG" || codon == "TGA") { // Stop codon found
                    cout << strand.substr(i, j + 3 - i) << endl;
                    found = true;
                    i = j + 2; // Move `i` to the end of this frame
                    break;
                }
                j += 3;
            }
        }
    }

    if (!found) {
        cout << "No reading frames found." << endl;
    }
}

int main(){
    
    int user_input = 0;
    string sequence1, sequence2;

    while (user_input != 7){

 
        cout << R"(--- DNA Analysis Menu ---
1. Calculate the similarity between two sequences of the same length
2. Calculate the best similarity between two sequences of either equal or unequal length
3. Identify mutations
4. Transcribe DNA to RNA
5. Find the reverse complement of a DNA sequence
6. Extract coding frames
7. Exit
Please enter your choice (1 - 7): )" << endl;
        cin >> user_input;


        switch (user_input)
        {
        case 1:
            cout << "Enter the first DNA sequence: " << endl;
            cin >> sequence1;
                while (!isValidStrand(sequence1)){
                    cout << "Invalid input. Please enter a valid sequence." << endl;
                    cout << "Enter the first DNA sequence: " << endl;
                    cin >> sequence1;
                }
            cout << "Enter the second DNA sequence: " << endl;
            cin >> sequence2;
                while (!isValidStrand(sequence2)){
                    cout << "Invalid input. Please enter a valid sequence." << endl;
                    cout << "Enter the second DNA sequence: " << endl;
                    cin >> sequence2;
                }
                if (strandSimilarity(sequence1, sequence2) == 0){
                    cout << "Error: Input strands must be of the same length." << endl;
                }else{
                    cout << "Similarity score: " << strandSimilarity(sequence1, sequence2) << endl;
                }
            break;
        case 2:
            cout << "Enter the first DNA sequence: " << endl;
            cin >> sequence1;
                while (!isValidStrand(sequence1)){
                    cout << "Invalid input. Please enter a valid sequence." << endl;
                    cout << "Enter the first DNA sequence: " << endl;
                    cin >> sequence1;
                }
            cout << "Enter the second DNA sequence: " << endl;
            cin >> sequence2;
                while (!isValidStrand(sequence2)){
                    cout << "Invalid input. Please enter a valid sequence." << endl;
                    cout << "Enter the second DNA sequence: " << endl;
                    cin >> sequence2;
                }

            bestStrandMatch(sequence1, sequence2);
            break;
        case 3:
            cout << "Enter the first DNA sequence: " << endl;
            cin >> sequence1;
                while (!isValidStrand(sequence1)){
                    cout << "Invalid input. Please enter a valid sequence." << endl;
                    cout << "Enter the first DNA sequence: " << endl;
                    cin >> sequence1;
                }
            cout << "Enter the second DNA sequence: " << endl;
            cin >> sequence2;
                while (!isValidStrand(sequence2)){
                    cout << "Invalid input. Please enter a valid sequence." << endl;
                    cout << "Enter the second DNA sequence: " << endl;
                    cin >> sequence2;
                }

            identifyMutations(sequence1, sequence2);
            break;
        case 4:
            cout << "Enter the DNA sequence to be transcribed:" << endl;
            cin >> sequence1;
                while (!isValidStrand(sequence1)){
                    cout << "Invalid input. Please enter a valid sequence." << endl;
                    cout << "Enter the DNA sequence to be transcribed: " << endl;
                    cin >> sequence1;
                }

            cout << "The transcribed DNA is: ";
            transcribeDNAtoRNA(sequence1);
            cout << endl;
            break;
        case 5:
            cout << "Enter the DNA sequence:" << endl;
            cin >> sequence1;
                while (!isValidStrand(sequence1)){
                    cout << "Invalid input. Please enter a valid sequence." << endl;
                    cout << "Enter the DNA sequence: " << endl;
                    cin >> sequence1;
                }

            cout << "The reverse complement is: ";
            reverseComplement(sequence1);
            cout << endl;
            break;
        case 6:
            cout << "Enter the DNA sequence: " << endl;
            cin >> sequence1;
            while (!isValidStrand(sequence1)){
                    cout << "Invalid input. Please enter a valid sequence." << endl;
                    cout << "Enter the DNA sequence: " << endl;
                    cin >> sequence1;
                }

            cout << "The extracted reading frames are: " << endl;
            getCodingFrames(sequence1);
            break;
        case 7:
            cout << "Exiting program." << endl;
            return 0; 
        default:
            cout << "Invalid input. Please select a valid option." << endl;
            break;
        }
    }
}