#!/usr/bin/perl -w
##############################################
### Huang He                               ###
### treatment_for_orignal_of_fasta         ###
##############################################

use Getopt::Long;
my %opt;
GetOptions(\%opt,"a=s","ac=s","f=s","am=s","ap=s","b=s","bml=s","bms=s","bv=s","c=s","help");
unless ((defined $opt{f}) or defined $opt{a} or defined $opt{am} or  defined $opt{ap} or defined $opt{ac} 
or defined $opt{b} or defined $opt{bml} or defined $opt{bms} or defined $opt{bv} 
or defined $opt{c} 
or defined $opt{help} ){
	&help (); 
}
&help()if(defined $opt{help});

if(defined $opt{a}){
	if(defined $opt{am} and defined $opt{ap} and defined $opt{ac}){
		`perl /home/huanghe/research/bin/pipeline/treatment_for_rRNA/trim_fasta.pl -f $opt{f} -m $opt{am} -p $opt{ap} -c $opt{ac}`;	
	}
	elsif(defined $opt{am} and defined $opt{ap} and !defined $opt{ac}){
		`perl /home/huanghe/research/bin/pipeline/treatment_for_rRNA/trim_fasta.pl -f $opt{f} -m $opt{am} -p $opt{ap}`;
	}
	elsif(defined $opt{am} and !defined $opt{ap} and defined $opt{ac}){
		`perl /home/huanghe/research/bin/pipeline/treatment_for_rRNA/trim_fasta.pl -f $opt{f} -m $opt{am} -c $opt{ac}`;
	}
	elsif(!defined $opt{am} and defined $opt{ap} and defined $opt{ac}){
		`perl /home/huanghe/research/bin/pipeline/treatment_for_rRNA/trim_fasta.pl -f $opt{f} -p $opt{ap} -c $opt{ac}`;
	}
	elsif(!defined $opt{am} and !defined $opt{ap} and defined $opt{ac}){
		`perl /home/huanghe/research/bin/pipeline/treatment_for_rRNA/trim_fasta.pl -f $opt{f} -c $opt{ac}`;
	}
	elsif(!defined $opt{am} and defined $opt{ap} and !defined $opt{ac}){
		`perl /home/huanghe/research/bin/pipeline/treatment_for_rRNA/trim_fasta.pl -f $opt{f} -p $opt{ap}`;
	}
	elsif(defined $opt{am} and !defined $opt{ap} and !defined $opt{ac}){
		`perl /home/huanghe/research/bin/pipeline/treatment_for_rRNA/trim_fasta.pl -f $opt{f} -m $opt{am}`;
	}
	elsif(!defined $opt{am} and !defined $opt{ap} and !defined $opt{ac}){
		`perl /home/huanghe/research/bin/pipeline/treatment_for_rRNA/trim_fasta.pl -f $opt{f}`;
	}
}

if(defined $opt{a}){
	if(defined $opt{b}){
		`/home/huanghe/research/tools/phred_cross/./cross_match.manyreads $opt{f}.salvaged $opt{bv} -minmatch $opt{bml} -minscore $opt{bms} -screen 2>cross_match.log`;
	}
}
else{
	if(defined $opt{b}){
		`/home/huanghe/research/tools/phred_cross/./cross_match.manyreads $opt{f} $opt{bv} -minmatch $opt{bml} -minscore $opt{bms} -screen 2>cross_match.log`;
	}
}

if(defined $opt{a} and defined $opt{b}){
	if(defined $opt{c}){
		open OUT,">batchfile" or die "can't batchfile";
		print OUT "unique.seqs(fasta=$opt{f}.salvaged.screen)";
		`/home/huanghe/research/tools/mothur/Mothur.cen_64/mothur/mothur batchfile`;
	}
}

elsif(!defined $opt{a} and defined $opt{b}){
	if(defined $opt{c}){
		open OUT,">batchfile" or die "can't batchfile";
		print OUT "unique.seqs(fasta=$opt{f}.screen)";
		`/home/huanghe/research/tools/mothur/Mothur.cen_64/mothur/mothur batchfile`;
	}
}

elsif(defined $opt{a} and !defined $opt{b}){
	if(defined $opt{c}){
		open OUT,">batchfile" or die "can't batchfile";
		print OUT "unique.seqs(fasta=$opt{f}.salvaged)";
		`/home/huanghe/research/tools/mothur/Mothur.cen_64/mothur/mothur batchfile`;
	}
}

elsif(!defined $opt{a} and !defined $opt{b}){
	if(defined $opt{c}){
		open OUT,">batchfile" or die "can't batchfile";
		print OUT "unique.seqs(fasta=$opt{f})";
		`/home/huanghe/research/tools/mothur/Mothur.cen_64/mothur/mothur batchfile`;
	}
}
	

sub help{
    warn <<"_EOT_"; 

******* This is a help document.********

	Usage: perl treatment_for_orignal_of_fasta.pl <options>
	
    	Options
    	
			-f	Input fasta_file key word (contains path), must be given
			
			*****************************************
			-a	Please input y to remove ploy tails.
			
			-am integer:	The minimum length allowed for a fasta record. (Optional. Default is 50.)
			
			-ap integer:	The minimum length of a Poly(A/T) sequence. (Optional. Default is 8.)
			
			-ac 1 or 0:	1 = Remove sequences with Poly(A/T) in the middle of the sequence, 0 = Don't remove. (Optional. Default is 0 = Don't remove.
			
			It returns two new fasta files: one with the removed sequences (fasta_file.removed), and one with the salvaged sequences (fasta_file.salvaged).
			
			*****************************************
			
			-b	Please input y to trim Vector Sequences.
			
			-bv	the file of vector sequences (fasta format; may just be one sequence).
			
			-bml suggested minimum length to match (10 is suggested). 
			
			-bms suggested minimum alignment score (10 is suggested).
			
			It returns a fasta.screen, which will contains the "vector masked" version of the original fasta file.
			
			***caution*** input fasta file is not aviliable for big size file(<200Mb).
			
			*****************************************
			 
			-c	Please input y to remove repeated sequences.
			
			It returns two new fasta files: one with the removed repeated sequences (fasta_file.unique.xxx), and one with the only name of sequences (fasta_file.names). 
			
			*****************************************
			
    	If input b, it will returns a record file: fasta_file.log
    	
    	If input c, it will returns a batchfile and a record file mothur.xxx.logfile	   
			

****************************************
_EOT_
    exit(1);
}