FreClu: Efficient Frequency-based De novo Short Read Clustering -- de novo clustering (last-modified : 20th August 2009)

This source code is bounded with test data of input, "testInputData_smallRNA.txt" with a tabular format, "Sequence \ Expectation of error number at each base \ Frequency".

Note: One has to compile the java files before running the program as below.
	javac MyTag_reverse.java
 	javac ClusteringOfShortTags_searchReverse_hash_open.java

Usage:
java ClusteringOfShortTags_searchReverse_hash_open -i filename

	REQUIRED 
		-i <filename of input>
	
	OPTIONAL 
		-o <filename of output; default is "clusterOutput.txt">
		-p <P-value threshold of parent-child detection: default is "0.01">
		-hashSize <bucket size of Hash table; default is "1,000,003".
			User may select other appropriate prime number for files of different size. 
			e.x.100,003; 3,000,017; 9,000,011>			
		-stat <table for converting Z-value to P-value. default is "normal_distribution.txt">
		

	OUTPUT FORMAT
		Because to generate all the trees needs too much I/O space, currently we only provide two output files named as "clusterOutput.txt" and "parent-childRelation.txt".
		"clusterOutput.txt":
			"Representative sequence \ Frequency of representative sequence \ Frequency of cluster \ Number of sequences in cluster \ Length of longest path".
		"parent-childRelation.txt":
			"Read sequence \ Representative sequence".

		Although in our program, we memorized all the sequences and their topologies in a cluster.
	
	
	EXAMPLE(Please try Java VM options -Xms and -Xmx to tune the heap size)
		java -Xms20G -Xmx20G ClusteringOfShortTags_searchReverse_hash_open -i testInputData_smallRNA.txt