Sequence File Handler Java Code 07-16-2014d1122

2015-01-13

fasta file handler
protein dna sequence file handler code
import fasta file sequences
export sequences to fasta file

code
{
import java.util.ArrayList;

public class SequenceFileHandler {

UsefulTools useful_tools = new UsefulTools();
	ArrayList names = new ArrayList();
	ArrayList sequences = new ArrayList();
	String current_line = "";
	
	
	public ArrayList getNames()
	{
		return names;
		
	}
	public ArrayList getSequences()
	{
		return sequences;
	}
	
	public ArrayList getSequencesFromFile(String file_path)
	{
		
		ArrayList lines = useful_tools.storeTextFiletoArrayList(file_path);
		boolean contains_greater_than_sign = false;
		for(int i=0; i<lines.size();i++)
		{
			String current_line = lines.get(i).toString();
			if(current_line.indexOf(">")!=-1)
			{
				contains_greater_than_sign = true;
			}
		}
		
		//if it contains a greater than sign, then this is a fasta file and the names and sequences should be collected accordingly
		//otherwise, each line without a space is a sequence, and the name should be some arbitrary name
		if(contains_greater_than_sign)
		{
			boolean title_found = false;
			
			for(int i=0; i<lines.size(); i++)
			{
				current_line = lines.get(i).toString();
				if(!title_found)
				{
					if(current_line.indexOf(">")!=-1 && current_line.length()>0)
					{
						title_found = true;
						names.add(current_line.replaceAll(">(.+)", "$1"));
					}
				}
				else if(title_found)
				{
					current_line = lines.get(i).toString();
					//make sure there are no spaces
					if(useful_tools.getMatchesRegEx(current_line, "\\s")==0 && current_line.length()>0)
					
					{
						
						sequences.add(current_line);
						title_found = false;
					}
				}
			}
		}
		else
		{
			int sequence_count = 0;
			//each line without a space should be a sequence
			for(int i=0; i<lines.size(); i++)
			{
				current_line = lines.get(i).toString();
				if(useful_tools.getMatchesRegEx(current_line, "\\s")==0 && current_line.length()>0)
				{
					
					sequences.add(current_line);
					sequence_count++;
					names.add("seq"+sequence_count);
				}
			}
		}
		return sequences;
	}
	
	/*
	 * this function expects sequences separated by commas
	 */
	public void createFastaFromSequences(String sequences, String directory, String filename)
	{
		createFastaFromSequences(useful_tools.stringWithCommasToArrayList(sequences), directory, filename);
	}
	
	public void createFastaFromSequences(ArrayList sequences, String directory, String filename)
	{
		String output_string = "";
		for(int i=0; i<sequences.size(); i++)
		{
			output_string+=">seq"+(i+1)+"\r\n"+sequences.get(i).toString()+"\r\n";
		}
		useful_tools.createTextFile(directory, filename, output_string);
	}
}

}

azim58wiki: