code editing convertSampleDataIntoCoIntensityNetwork function 11-06-2014d1616

2014-11-07

code editing convertSampleDataIntoCoIntensityNetwork function 11-06-2014d1616

11-07-2014d1327

I think using all 45,000 genes from some of the gene expression datasets will produce networks that are too big to analyze with a normal analysis. I think I will just randomly select 500 genes.

11-07-2014d1011

I'll try to precalculate parts for lists.

formula for pearsons correlation coefficient here

http://www.socscistatistics.com/tests/pearson/

Is Sum((xi-xbar)*(yi-ybar)) the same as Sum(xi-xbar)*Sum(yi-ybar)?

I'll do a little test with a few lists.

"C:\Sync\pDR\2014\2014\11-07-2014d1014\Calculation test 11-07-2014d1014.xlsx"
-This spreadsheet also has a spreadsheet calculating later verifcations of the correlation calculation.

No they are not the same thing. Sum(xi-xbar)*Sum(yi-ybar) will always be 0.

code before trying to split matrix into blocks



	public void convertSampleDataIntoCoIntensityNetwork(double[][] values, double threshold, String output_directory)


	{


		PearsonsCorrelation pc = new PearsonsCorrelation();


		


		String[] sarray = new String[values.length];


		useful_tools.createTextFile(output_directory, "row1.txt", Arrays.toString(values[0]));


		//fill up the similarity score with absolute value correlation coefficient values, but just write each row to a separate text file


		


		String edges = "";


		for(int i=0; i<values.length; i++)


		{


			System.out.println(i);


			String current_row = "";


			for(int j=i+1; j<values.length; j++)


			{


				


				double calculated_similarity_score = Math.abs(pc.correlation(values[i], values[j]));


				int link=0;


				if(calculated_similarity_score>threshold)


				{


					link = 1;

if(i==(values.length-1) && link==1)



					edges+=i+"\t"+j;

else if(link==1)



					edges+=i+"\t"+j+"\r\n";

}
}

useful_tools.createTextFile(output_directory, "network.txt", edges);
}
}