Add option to set minimal size of sequences to use.

code/reas_all.pl

@@ -25,6 +25,7 @@ Usage : $0 [options]
 	-output	STRING	output final consensus file, default="consensus.fa"
 	-prefix	STRING	prefix of consensus sequence name, default=""
 	-log		output log details, default=[not]
+	-seqsize INT minimal length of input sequences to use
 
 
 Advanced options:
@@ -47,7 +48,7 @@ die $usage if (@ARGV<1);
 
 my %opts;
 GetOptions(\%opts, "read=s", "k=i", "d=i", "m=i", "subset=s", "n=i", "pa=i", "bound=s", "link=s", "multi=i", "output=s",
-"prefix:s", "log", "size=i", "ident=s", "nonlcs=i", "end=i", "min_depth=i", "min_extend=i", "max_extend=i", "h");
+"prefix:s", "log", "size=i", "ident=s", "nonlcs=i", "end=i", "min_depth=i", "min_extend=i", "max_extend=i", "seqsize=i", "h");
 
 die $usage if defined $opts{h};
 my $read_file = defined $opts{read}? $opts{read} : die();
@@ -64,6 +65,7 @@ my $output = defined $opts{output}? $opts{output} : "consensus.fa";
 my $prefix = defined $opts{prefix}? "-p $opts{prefix}" : "";
 my $log = defined $opts{log}? "-s" : "";
 my $size = defined $opts{size}? $opts{size} : 100;
+my $seqsize = defined $opts{seqsize}? $opts{seqsize} : 100;
 my $ident = defined $opts{ident}? $opts{ident} : 0.6;
 my $nonlcs = defined $opts{nonlcs}? $opts{nonlcs} : 50;
 my $end = defined $opts{end}? $opts{end} : 50;
@@ -76,7 +78,7 @@ my $nmers_file = "read.N_mers";
 my $hd_read_id_file = "HD_reads.id";
 my $hd_read_file = "HD_reads.fa";
 my $seg_file = "seg.fa";
-`cat $read_file |CleanData.pl |rename.pl >$clean_read_file`;
+`cat $read_file |CleanData.pl -size=$seqsize |rename.pl >$clean_read_file`;
 `kmer_num -k $k $clean_read_file| kmer2reads -d $d $clean_read_file |N_mers -m $m >$nmers_file`;
 `cat $nmers_file |cut -f1 |RandomList.pl $fraction >$hd_read_id_file`;
 `cat $clean_read_file |pickListSeq.pl $hd_read_id_file >$hd_read_file`;