MSAEvaluator

package gbi;

import java.io.*;

/**
 * Reads in an MSA in FastA format and a BLOSUM matrix and then computes the alignment score
 * Exercise 4, Assignment 5
 *
 * @author Famke Bäuerle and Viktor Schiff, Code by Kay Nieselt and Daniel Huson, 5.2017
 */
public class MSAEvaluator {
	private FastA msa; // the input file
	private BlosumMatrix blosumMatrix; // the BLOSUM matrix

	/**
	 * Constructor
	 *
	 * @param fastAFile file containing the MSA
	 * @param blosumFile file containing the BLOSUM scoring blosumMatrix
	 * @throws IOException
	 */
	public MSAEvaluator (String fastAFile,String blosumFile) throws IOException {
		// initialize the members msa and blosumMatrix members and read in the data
		msa = new FastA();
		BufferedReader r1 = new BufferedReader(new FileReader(fastAFile));
		msa.read(r1);
		blosumMatrix = new BlosumMatrix();
		BufferedReader r2 = new BufferedReader(new FileReader(blosumFile));
		blosumMatrix.read(r2);

	}

	/**
	 * check whether the msa member contains a valid MSA (i.e., are all the sequences of the same length)
	 *
	 * @return true, if variable msa contains a valid MSA
	 */
	public boolean isValidMSA() {
		int numSeq = msa.size();
		String[] allSeq = new String[numSeq];
		for (int i = 0; i < numSeq; i++) { //put all Sequences into array
			allSeq[i] = msa.getSequence(i);
		}
		int[] seqLen = new int[numSeq];
		for (int i = 0; i < numSeq; i++) { //get length of every sequence
			seqLen[i] = allSeq[i].length();
		}
		for(int i = 1; i < seqLen.length; i++) { //compare all lengths with first length
			if(seqLen[0] != seqLen[i]) return false; //if length is different, return false
		}
		return true;
	}

	/**
	 * computes the sum of pairs score for the alignment. Use the last row and column of the blosum matrix - denoted by * - as gap penality
	 *
	 * @return sum of pairs score
	 */
	public int computeSumOfPairsScore () {
		int rowScore = 0;
		int score = 0;
		char[] rowChars = new char[msa.size()];
		for (int j = 0; j < msa.getSequence(0).length(); j++) { //loop through the whole sequence
			for (int i = 0; i < msa.size(); i++) { //loop through all sequences
				rowChars[i] = msa.getSequence(i).charAt(j); //fill char array with char from every sequence at the same index
				//loop through all chars and compare them
				for (int p = 0; p < rowChars.length - 1; p++) { //0 to n-1
					for (int k = p + 1; k < rowChars.length; k++) { //1 to n
						char char1 = rowChars[p];
						char char2 = rowChars[k];
						if (char1 == char2 && char1 == '-') { //both char1 and char2 are gaps
							rowScore = rowScore + 0;
						}
						else if (char1 == '-' || char2 == '-') { //either char1 or char 2 is gap
							rowScore = rowScore - 4;
						}
						else { //char1 and char2 are amino acids
							rowScore = rowScore + blosumMatrix.getScore(char1, char2);
						}
					}
				}
				score = score + rowScore;
				rowScore = 0;
			}

		}
		//compare every Sequence with the other sequences
		/* for (int i = 0; i < msa.size() - 1; i++) { //0 to n-1
			String seq1 = msa.getSequence(i);
			for (int j = i + 1; j < msa.size(); j++) { //1 to n
				String seq2 = msa.getSequence(j);
				//loop through all letters of both Sequences
				for (int p = 0; p < seq1.length(); p++) {
					char char1 = seq1.charAt(p);
					char char2 = seq2.charAt(p);
					if (char1 == char2 && char1 == '-') { //both char1 and char2 are gaps
						score = score + 0;
					}
					else if (char1 == '-' || char2 == '-') { //either char1 or char 2 is gap
						score = score - 4;
					}
					else { //char1 and char2 are amino acids
						score = score + blosumMatrix.getScore(char1, char2);
					}
				}
			}
		} */

		return score;
	}

	/**
	 * gets the alignment
	 *
	 * @return MSA
	 */
	public FastA getMsa() {
		return msa;
	}

	/**
	 * gets the BLOSUM matrix
	 *
	 * @return BLOSUM matrix
	 */
	public BlosumMatrix getBlosumMatrix() {
		return blosumMatrix;
	}

	/**
	 * run the program
	 *
	 * @param args
	 * @throws IOException
	 */
	public static void main (String[] args) throws IOException {
		System.out.println("Famke Bäuerle and Viktor Schiff");

		MSAEvaluator msaEvaluator=new MSAEvaluator("msa.fasta","blosum62.txt");

		//msaEvaluator.getMsa().write(new OutputStreamWriter(System.out));
		msaEvaluator.getBlosumMatrix().write(new OutputStreamWriter(System.out));

		if(msaEvaluator.isValidMSA()) {
			System.out.println("Sum of pairs score:  " + msaEvaluator.computeSumOfPairsScore());
		}
		else
			System.err.println("Alignment invalid");
	}
}