% Loads matrix Data.m with sequences in binary MATLAB format.
% Writes out the data in fastA and interleaved Phylip format.

clear
load Data
K=size(Data,1)
N=size(Data,2)
D=size(Data,3)

d= input('Which data set do you want to retrieve?');

fid =fopen('dna.fastA','w');
for k=1:K
	strain=strcat('>strain',num2str(k));
	fprintf(fid,'%8s\n',strain);
	for n=1:N
		if Data(k,n,d)==1
			fprintf(fid,'%1s','A');
		elseif Data(k,n,d)==2
			fprintf(fid,'%1s','C');
		elseif Data(k,n,d)==3
			fprintf(fid,'%1s','G');
		elseif Data(k,n,d)==4
			fprintf(fid,'%1s','T');
		else
			error('Unknown nucleotide');
		end
		if mod(n,50)==0 | n==N
			fprintf(fid,'\n');
		end
	end
end  
fclose(fid);

% Get FastA format into sequential Phylip format
!readseq -f=Phylip3.2 dna.fastA -out=dna.phyB -a
!sed 's/ YF//g' dna.phyB > stercus
!/bin/mv stercus dna.phyB