FreClu: Efficient Frequency-based De novo Short Read Clustering -- preparation for input data


Please try Java VM options -Xms and -Xmx to tune the heap size.
E.g. java -Xms10G -Xmx10G Radix_DNAseq <input file>
Note: One has to compile the java files (javac) before running the program below.
	javac QV_format1.java
	javac QV_format1_qseq.java
	javac QV_format1_fastq.java
	javac QV_filter.java
	javac QV_format2.java
	javac ChangeSolexaQVCountTagToFASTA.java
	javac OverlapWithoutGap.java
	javac AlignmentMain.java
	javac FormatOverlap.java
	javac Merge_randomModel.java
	javac Radix_DNAseq.java
	javac Merge.java
	javac Radix_num.java

Usage:
Five steps are included as below.

(1) REQUIRED : 
	(1)-1. To join all the raw sequence files Illumina <*_seq.txt > and their QV files Illumina <*_prb.txt >, and remove sequence which has ambiguous base N. The output file will be named as <*_seq-prb.txt>. 
	java QV_format1 <*_seq.txt> 

	OR
	(1)-2. For Illumina <*qseq.txt> format raw sequence files which have merged sequences and QVs.
	java QV_format1_qseq <*_qseq.txt> 

	OR
	(1)-3. For Illumina <*fastq> format raw sequence files which have merged sequences and QVs.
	java QV_format1_fastq <*fastq> 
 
(2) OPTIONAL : For output file <*_seq-prb.txt> of (1), set a QV filter to trim apparently low quality reads; What we used was, at most 4 of the first 20 bases of a read were allowed to have QV < 9.
	java QV_filter <output file of (1)> 9 4 20 

(3) REQUIRED : For the output file of (2), cut 3' linker sequence.
	(3)-1. For 5'-end SAGE sample, we trimmed all reads at 25 bases. (If it is not necessary to cut any 3' linker sequence, please try "java QV_format2 <output file of (2)>". )
		java QV_format2 <output file of (2)> 25 
		
	OR
	(3)-2. For small RNAs sample, we trimmed 3' linker that ovelaped to the linker "TCGTATGCCGTCTTCTGCTTGT" at 3' end with at least 5 bases and over 80 percent similarity if alignment length is less than 11 bases. The output file named "*_format_matchToLinker" should be used for further analysis. Other output files, which are "*_format_linkerAtStart" and "*_format_unmatchToLinker" consist of short reads failed in 3'linker mapping.
		(3)-2-1. java ChangeSolexaQVCountTagToFASTA <output file of (2)> 
		(3)-2-2. java AlignmentMain <output file of (3)-2-1> 
		(3)-2-3. java FormatOverlap <output file of (3)-2-2>  5 11 0.8 
		
(4) REQUIRED : 
	(4)-1. Sort redundan reads in the output file of (3) in lexicographical order.
		java Radix_DNAseq <output file of (3)> 
	(4)-2. Scan the output file of (4)-1 to count the frequency of individual non-redundant sequence and the sum of expected errors at each base (NOTE: Illumina QV). 
		(4)-2-1. For data with empirical BAC-based DNA substitution file<trans_corres> and adjusted QV file <qv_corres>, 
			java Merge <output file of (4)-1> -q <qv_corres> -s <trans_corres>
			
		OR 
		(4)-2-2. For data without files above, an evenly distributed DNA substitution pattern (1/3) and non-adjusted QV are used. 
			java Merge_randomModel <output file of (4)-1> 
			
(5) REQUIRED : Sort non-redundant sequences the output file of (4)-2 according to their frequencies in descending order using the radix-sort algorithm.
	java Radix_num  <output file of (4)-2>