diff --git a/bin/homolog_genewise b/bin/homolog_genewise
index d82522a..fe99d75 100755
--- a/bin/homolog_genewise
+++ b/bin/homolog_genewise
@@ -149,7 +149,7 @@ unless (-e "2.blast.ok") {
         if (($segmentSize - $_[9] + 1) < 1000) { $keep = 0; }
 
         $_[8] += $pos;
-        $_[9] += + $pos;
+        $_[9] += $pos;
         print OUT (join "\t", @_);
     }
     close OUT;
diff --git a/bin/homolog_predict.pl b/bin/homolog_predict.pl
deleted file mode 100755
index 406513b..0000000
--- a/bin/homolog_predict.pl
+++ /dev/null
@@ -1,170 +0,0 @@
-#!/usr/bin/env perl
-
-use strict;
-use Getopt::Long;
-use Cwd qw/abs_path getcwd cwd/;
-use File::Basename;
-
-my $usage = <<USAGE;
-Usage:
-    $0 [options] homolog_proteins.fasta genome_seq.fasta > out.gff
-
-    --cpu <int>    default: 8
-    设置diamond程序使用线程数，genewise或gth命令运行的并行数。
-
-    --coverage_ratio <float>    default: 0.4
-    对diamond blastx的结果进行过滤，要求对同源蛋白的覆盖率不小于该值。
-
-    --evalue <float>    default: 1e-9
-    对diamond blastx的结果进行过滤，要求对同源蛋白的evalue值不大于该值。
-
-    --method <string>    default: gth
-    设置使用同源蛋白进行基因预测的方法，可设置值为gth或genewise。一般情况下，genewise进行基因预测的耗时是gth的五倍。
-
-    --tmp_dir <string>    default: tmp_\$date\$pid
-    程序运行时临时文件夹名称。
-
-    --help    default: None
-    display this help and exit.
-
-USAGE
-if (@ARGV==0){die $usage}
-
-my ($cpu, $coverage_ratio, $evalue, $method, $tmp_dir, $help_flag);
-GetOptions(
-    "cpu:i" => \$cpu,
-    "coverage_ratio:f" => \$coverage_ratio,
-    "evalue:f" => \$evalue,
-    "method:s" => \$method,
-    "tmp_dir:s" => \$tmp_dir,
-    "help" => \$help_flag,
-);
-$cpu ||= 8;
-$coverage_ratio ||= 0.4;
-$evalue ||= 1e-9;
-$method ||= "gth";
-unless ($method eq "gth" or $method eq "genewise") {
-    die "--method was set wrong, it should be set to gth or genewise. \n";
-}
-my $date = `date +%Y%m%d%H%M%S`; chomp($date);
-$tmp_dir ||= "tmp_$date$$";
-$tmp_dir = abs_path($tmp_dir);
-mkdir $tmp_dir unless -e $tmp_dir;
-
-if ( $help_flag ) { die $usage }
-
-my $input_protein = abs_path($ARGV[0]);
-my $input_genome = abs_path($ARGV[1]);
-
-my $pwd = `pwd`; print STDERR "##########\nPWD (Current Directory): $pwd";
-print STDERR (localtime) . "CMD (Main Program): $0 " . join(" ", @ARGV) . "\n##########\n\n";
-
-
-# 1. 运行diamond分析，寻找蛋白序列和参考基因组的匹配位点。
-print STDERR "1. align genome sequences against homolog proteins through diamond software.\n";
-# 1.1 运行 diamond makedb 命令，将蛋白序列做成数据库。
-print STDERR "1.1 run diamond makedb, get a protein database.\n";
-chdir $tmp_dir;
-my $pwd = `pwd`; print STDERR "PWD: $pwd";
-my $cmdString = "diamond makedb --threads $cpu --db homolog --in $input_protein 2> 1.1.diamond_makedb.log";
-unless (-e "1.1.diamond_makedb.ok") {
-	print STDERR (localtime) . ": CMD: $cmdString\n";
-	system("$cmdString") == 0 or die "failed to execute: $cmdString\n";
-	open OUT, ">", "1.1.diamond_makedb.ok" or die $!; close OUT;
-}
-else {
-	print STDERR "CMD(Skipped): $cmdString\n";
-}
-
-# 1.2 运行 diamond blastx 命令将基因组序列比对到蛋白序列数据库上。注意 --max-target-seqs 参数设置为蛋白序列的数量。
-print STDERR "1.2 split genome sequences, and get diamond blastx commands.\n";
-unless ( -e "1.2.split_genome_seqs_and_get_cmds.ok" ) {
-	# 获取蛋白序列数量
-	open IN, $input_protein or die "Can not open file $input_protein, $!";
-	my $protein_number = 0;
-	while (<IN>) {
-		$protein_number ++ if m/^>/;
-	}
-	close IN;
-	# 将基因组序列分割成单独的序列，得到diamond blastx命令
-	open IN, $input_genome or die "Can not open file $input_genome, !";
-	my (%seq, $seq_ID);
-	while (<IN>) {
-		if (m/^>(\S+)/) {
-			$seq_ID = $1;
-		}
-		else {
-			chomp;
-			$seq{$seq_ID} .= $_;
-		}
-	}
-	close IN;
-	mkdir "split_genome_seqs" unless -e "split_genome_seqs";
-	open CMD1, ">", "split_genome_seqs/command.diamond_blastx.list" or die "Can not create file split_genome_seq/command.diamond_blastx.list, $!";
-	open CMD2, ">", "split_genome_seqs/command.parsing_blast_result.list" or die "Can not create file split_genome_seq/command.parsing_blast_result.list, $!";
-	foreach ( keys %seq ) {
-		open OUT, ">", "split_genome_seqs/$_.fa" or die "Can not create file split_genome_seq/$_.fa, $!";
-		print OUT ">$_\n$seq{$_}\n";
-		close OUT;
-		print CMD1 "diamond blastx --db homolog --query split_genome_seqs/$_.fa --out split_genome_seqs/$_.xml --outfmt 5 --sensitive --max-target-seqs $protein_number --evalue 1e-3 --id 10 2> split_genome_seqs/$_.log\n";
-		print CMD2 "parsing_blast_result.pl --type xml --no-header --max-hit-num $protein_number --evalue $evalue --subject-coverage $coverage_ratio --query-coverage 0 split_genome_seqs/$_.xml > split_genome_seqs/$_.tab\n";
-	}
-	close CMD1; close CMD2;
-	open OUT, ">", "1.2.split_genome_seqs_and_get_cmds.ok" or die $!; close OUT;
-}
-else {
-	print STDERR "This step was skipped, for the file 1.2.split_genome_seqs_and_get_cmds.ok exists.\n";
-}
-
-# 1.3 并行化运行 diamond blastx 命令
-# 按内存余量确定 diamond 的并行化数量。按单个 dianmond 命令峰值消耗 60 Gb 内存。
-my $MemAvailable = &get_MemAvailable();
-my $paraFly_CPU = 1;
-$paraFly_CPU = $cpu / 4 if $paraFly_CPU < ($cpu / 4);
-$paraFly_CPU = $MemAvailable / 60000000 if $paraFly_CPU > ($MemAvailable / 60000000);
-$paraFly_CPU = 1 if $paraFly_CPU < 1;
-my $cmdString = "ParaFly -c split_genome_seqs/command.diamond_blastx.list -CPU $paraFly_CPU 2> 1.3.ParaFly_diamond_blastx.log";
-unless ( -e "1.3.ParaFly_diamond_blastx.ok" ) {
-	print STDERR (localtime) . ": CMD: $cmdString\n";
-	if ( system("$cmdString") != 0 ) {
-		$cmdString = "ParaFly -c split_genome_seqs/command.diamond_blastx.list -CPU 1 2>> 1.3.ParaFly_diamond_blastx.log";
-		print STDERR (localtime) . ": CMD: $cmdString\n";
-		system("$cmdString") == 0 or die "failed to execute: $cmdString\n";
-	}
-	open OUT, ">", "1.3.ParaFly_diamond_blastx.ok" or die $!; close OUT;
-}
-else {
-	print STDERR "CMD(Skipped): $cmdString\n";
-}
-
-# 1.4 解析 xml 文件，得到基因组区域及其匹配的氨基酸序列名称。
-$cmdString = "ParaFly -c split_genome_seqs/command.parsing_blast_result.list -CPU $cpu 2> 1.4.ParaFly_parsing_blast_result.log";
-unless ( -e "1.4.ParaFly_parsing_blast_result.ok" ) {
-	print STDERR (localtime) . ": CMD: $cmdString\n";
-	if ( system("$cmdString") != 0 ) {
-		$cmdString = "ParaFly -c split_genome_seqs/command.parsing_blast_result.list -CPU 1 2> 1.4.ParaFly_parsing_blast_result.log";
-		print STDERR (localtime) . ": CMD: $cmdString\n";
-		system("$cmdString") == 0 or die "failed to execute: $cmdString\n";
-	}
-	open OUT, ">", "1.4.ParaFly_parsing_blast_result.ok" or die $!; close OUT;
-}
-else {
-	print STDERR "CMD(Skipped): $cmdString\n";
-}
-
-# 2. 解析diamond比对结果，
-# 2.1 根据同源蛋白的比对结果，推测出物种的基因长度信息。
-print STDERR "2
-
-sub get_MemAvailable {
-    open IN, "/proc/meminfo" or die "Can not open file /proc/meminfo, $!";
-    my $MemAvailable;
-    while (<IN>) {
-        if (m/MemAvailable:\s*(\d+)\s*kB/) {
-            $MemAvailable = $1;
-            next;
-        }
-    }
-    close IN;
-    return $MemAvailable;
-}
diff --git a/bin/homolog_prediction b/bin/homolog_prediction
new file mode 100755
index 0000000..c2becb7
--- /dev/null
+++ b/bin/homolog_prediction
@@ -0,0 +1,104 @@
+#!/usr/bin/env perl
+
+use strict;
+use Getopt::Long;
+use Cwd qw/abs_path getcwd cwd/;
+use File::Basename;
+
+my $binPath = dirname($0);
+
+my $usage = <<USAGE;
+Usage:
+    $0 [options] homolog_proteins.fasta genome_seq.fasta > out.gff
+
+	--out_prefix <string>    default: out
+	设置输出文件前缀。程序默认输出如下文件：（1）out.diamond.tab，将基因组序列和同源蛋白进行dimond blastx比对的结果文件；（2）out.
+
+    --segmentSize <int>    default: 1000000
+    --overlapSize <int>    default: 100000
+    程序将基因组较长的序列进行分割，以加快diamond blastx的并行化比对速度；若单条序列长度超过1Mb， 则将单条序列进行切割，两条相邻的序列间重叠的长度为100kb。
+
+    --cpu <int>    default: 8
+    设置diamond程序使用线程数，genewise或gth命令运行的并行数。
+
+	--identity <float>    default: 0.2
+    --evalue <float>    default: 1e-9
+    设置diamond blastx分析时的Identity和E-value阈值。由于diamond blastx进行分析，每个Hit仅给出一个最优的HSP（而NCBI Blast+会给出所有HSPs），因此不需要根据覆盖率来对BLASTX结果进行过滤。
+
+    --method <string>    default: gth
+    设置使用同源蛋白进行基因预测的方法，可设置值为gth或genewise。一般情况下，genewise进行基因预测的耗时是gth的五倍。
+
+    --tmp_dir <string>    default: tmp_\$date\$pid
+    程序运行时临时文件夹名称。
+
+    --help    default: None
+    display this help and exit.
+
+USAGE
+if (@ARGV==0){die $usage}
+
+my ($out_prefix, $segmentSize, $overlapSize, $cpu, $identity, $evalue, $method, $tmp_dir, $help_flag);
+GetOptions(
+	"out_prefix:s" => \$out_prefix,
+    "segmentSize:i" => \$segmentSize,
+    "overlapSize:i" => \$overlapSize,
+    "cpu:i" => \$cpu,
+	"identity:f" => \$identity,
+    "evalue:f" => \$evalue,
+    "method:s" => \$method,
+    "tmp_dir:s" => \$tmp_dir,
+    "help" => \$help_flag,
+);
+$out_prefix ||= "out";
+$segmentSize ||= 1000000;
+$overlapSize ||= 100000;
+$cpu ||= 8;
+$identity ||= 0.2;
+$evalue ||= 1e-9;
+$method ||= "gth";
+unless ($method eq "gth" or $method eq "genewise") {
+    die "--method was set wrong, it should be set to gth or genewise. \n";
+}
+my $date = `date +%Y%m%d%H%M%S`; chomp($date);
+$tmp_dir ||= "tmp_$date$$";
+$tmp_dir = abs_path($tmp_dir);
+mkdir $tmp_dir unless -e $tmp_dir;
+
+if ( $help_flag ) { die $usage }
+
+my $input_protein = abs_path($ARGV[0]);
+my $input_genome = abs_path($ARGV[1]);
+
+my $pwd = `pwd`; print STDERR "##########\nPWD (Current Directory): $pwd";
+print STDERR (localtime) . "CMD (Main Program): $0 " . join(" ", @ARGV) . "\n##########\n\n";
+
+
+# 1. 运行diamond分析，寻找蛋白序列和参考基因组的匹配位点。
+print STDERR "1. align genome sequences against homolog proteins through diamond software.\n";
+chdir $tmp_dir;
+my $pwd = `pwd`; print STDERR "PWD: $pwd";
+my $cmdString = "$binPath/homolog_prediction.01paraDiamond --segmentSize $segmentSize --overlapSize $overlapSize --cpu $cpu --identity $identity --evalue $evalue --tmp_dir a.para_diamond $input_protein $input_genome > $out_prefix.diamond.tab 2> a.para_diamond.log";
+unless (-e "1.paraDiamond.ok") {
+    print STDERR (localtime) . ": CMD: $cmdString\n";
+    system("$cmdString") == 0 or die "failed to execute: $cmdString\n";
+    open OUT, ">", "1.paraDiamond.ok" or die $!; close OUT;
+}
+else {
+	print STDERR "CMD(Skipped): $cmdString\n";
+}
+
+# 2. 过滤被包含匹配区域中得分过低的比对结果。
+print STDERR "2. filtering the low quality alignments when contained in another match region.\n";
+my $cmdString = "$binPath/homolog_prediction.02filtringLowQualityHits ";
+unless (-e "2.filtringLowQualityHits.ok") {
+    print STDERR (localtime) . ": CMD: $cmdString\n";
+    system("$cmdString") == 0 or die "failed to execute: $cmdString\n";
+    open OUT, ">", "2.filtringLowQualityHits.ok" or die $!; close OUT;
+}
+else {
+	print STDERR "CMD(Skipped): $cmdString\n";
+}
+
+# 3. 调用genewise(wise)进行基因预测。
+# 4. 调用gth(genomethreader)进行基因预测。
+# 5. 合并所有的基因预测结果，若一个区域有多个蛋白预测的结果，取其一致性结果。
diff --git a/bin/homolog_prediction.01paraDiamond b/bin/homolog_prediction.01paraDiamond
new file mode 100755
index 0000000..339f3fb
--- /dev/null
+++ b/bin/homolog_prediction.01paraDiamond
@@ -0,0 +1,195 @@
+#!/usr/bin/env perl
+
+use strict;
+use Getopt::Long;
+use Cwd qw/abs_path getcwd cwd/;
+use File::Basename;
+
+my $usage = <<USAGE;
+Usage:
+    $0 [options] homolog_proteins.fasta genome_seq.fasta > out.tab
+
+    本程序用于将全基因组序列和临近物种的同源蛋白使用diamond blastx进行比对，得到表格结果。若使用dianmond命令直接对单个全基因组Fasta文件进行分析，单个diamond命令并不能一直占用单台服务器全部CPU线程，有很长时间的单线程运行阶段，从而导致程序运行速度较慢。因此，推荐使用本程序将基因组序列进行分割后并行化计算，再合并结果，从而加快计算速度。
+
+    --segmentSize <int>    default: 1000000
+    --overlapSize <int>    default: 100000
+    程序将基因组较长的序列进行分割，以加快diamond blastx的并行化比对速度；若单条序列长度超过1Mb， 则将单条序列进行切割，两条相邻的序列间重叠的长度为100kb。
+
+    --cpu <int>    default: 8
+    设置CPU线程数。本程序调用diamond命令时未设置--threads参数，表明单个diamond默认使用所有CPU线程；但diamond程序后半部分往往对CPU线程利用较少，往往只能使用4个CPU线程，因此本程序并行化运行diamond的数量为--cpu参数设置的值除以4；此外，考虑到diamond比较消耗内存，单个diamond程序消耗内存峰值10G，并行化运行diamond的数量最高为剩余内存量除以10G；当调用ParaFly并行化运行失败后，程序会使用1个并行数来运行diamond命令。
+
+    --identity <float>    default: 0.2
+    --evalue <float>    default: 1e-9
+    设置diamond blastx分析时的Identity和E-value阈值。由于diamond blastx进行分析，每个Hit仅给出一个最优的HSP（而NCBI Blast+会给出所有HSPs），因此不需要根据覆盖率来对BLASTX结果进行过滤。
+
+    --tmp_dir <string>    default: tmp_\$date\$pid
+    程序运行时临时文件夹名称。
+
+    --help    default: None
+    display this help and exit.
+
+USAGE
+if (@ARGV==0){die $usage}
+
+my ($segmentSize, $overlapSize, $cpu, $identity, $evalue, $tmp_dir, $help_flag);
+GetOptions(
+    "segmentSize:i" => \$segmentSize,
+    "overlapSize:i" => \$overlapSize,
+    "cpu:i" => \$cpu,
+    "identity:f" => \$identity,
+    "evalue:f" => \$evalue,
+    "tmp_dir:s" => \$tmp_dir,
+    "help" => \$help_flag,
+);
+$segmentSize ||= 1000000;
+$overlapSize ||= 100000;
+$cpu ||= 8;
+$identity ||= 0.2;
+$evalue ||= 1e-9;
+my $date = `date +%Y%m%d%H%M%S`; chomp($date);
+$tmp_dir ||= "tmp_$date$$";
+$tmp_dir = abs_path($tmp_dir);
+mkdir $tmp_dir unless -e $tmp_dir;
+
+if ( $help_flag ) { die $usage }
+
+my $input_protein = abs_path($ARGV[0]);
+my $input_genome = abs_path($ARGV[1]);
+
+my $pwd = `pwd`; print STDERR "##########\nPWD (Current Directory): $pwd";
+print STDERR (localtime) . "CMD (Main Program): $0 " . join(" ", @ARGV) . "\n##########\n\n";
+
+
+# 1 运行 diamond makedb 命令，将蛋白序列做成数据库。
+print STDERR "1. run diamond makedb, get a protein database.\n";
+chdir $tmp_dir;
+my $pwd = `pwd`; print STDERR "PWD: $pwd";
+my $cmdString = "diamond makedb --threads $cpu --db homolog --in $input_protein 2> 1.diamond_makedb.log";
+unless (-e "1.diamond_makedb.ok") {
+    print STDERR (localtime) . ": CMD: $cmdString\n";
+    system("$cmdString") == 0 or die "failed to execute: $cmdString\n";
+    open OUT, ">", "1.diamond_makedb.ok" or die $!; close OUT;
+}
+else {
+    print STDERR "CMD(Skipped): $cmdString\n";
+}
+
+# 2 运行 diamond blastx 命令将基因组序列比对到蛋白序列数据库上。注意 --max-target-seqs 参数设置为蛋白序列的数量。
+print STDERR "2. split genome sequences, and get diamond blastx commands.\n";
+unless ( -e "2.split_genome_seqs_and_get_cmds.ok" ) {
+    # 获取蛋白序列数量
+    open IN, $input_protein or die "Can not open file $input_protein, $!";
+    my $protein_number = 0;
+    while (<IN>) {
+        $protein_number ++ if m/^>/;
+    }
+    close IN;
+    # 将基因组序列按长度进行分割，得到diamond blastx命令
+    open IN, $input_genome or die "Can not open file $input_genome, !";
+    my (%seq, $seq_ID);
+    while (<IN>) {
+        chomp;
+        if (m/^>(\S+)/) { $seq_ID = $1; }
+        else { $seq{$seq_ID} .= $_; }
+    }
+    close IN;
+    mkdir "split_genome_seqs" unless -e "split_genome_seqs";
+    open CMD, ">", "split_genome_seqs/command.diamond_blastx.list" or die "Can not create file split_genome_seq/command.diamond_blastx.list, $!";
+    my @seq_ID;
+    foreach my $seq_id ( sort keys %seq ) {
+        my $seq = $seq{$seq_id};
+        my $seq_length = length $seq;
+        if ($seq_length > $segmentSize) {
+            my @partion = &get_partition($seq_length, $segmentSize, $overlapSize);
+            foreach (@partion) {
+                my $start = $_ - 1;
+                my $end = $start + $segmentSize;
+                my $sub_seq = substr($seq, $start, $segmentSize);
+                open OUT, ">", "split_genome_seqs/$seq_id.$start.fa" or die "Can not create file split_genome_seqs/$seq_id.$start.fa, $!\n";
+                print OUT ">$seq_id\_start_$start\n$sub_seq\n";
+                close OUT;
+                push @seq_ID, "$seq_id.$start";
+            }
+        }
+        else {
+            open OUT, ">", "split_genome_seqs/$seq_id.0.fa" or die "Can not create file split_genome_seqs/$seq_id.0.fa, $!\n";
+            print OUT ">$seq_id\_start_0\n$seq\n";
+            close OUT;
+            push @seq_ID, "$seq_id.0";
+        }
+    }
+    foreach ( @seq_ID ) {
+        print CMD "diamond blastx --db homolog --query split_genome_seqs/$_.fa --out split_genome_seqs/$_.out --outfmt 6 --sensitive --max-target-seqs $protein_number --evalue $evalue --id $identity 2> split_genome_seqs/$_.log\n";
+    }
+    close CMD;
+    open OUT, ">", "2.split_genome_seqs_and_get_cmds.ok" or die $!; close OUT;
+    my $sequence_number = @seq_ID;
+    print STDERR "The genome was split to $sequence_number sequences.\n";
+}
+else {
+    print STDERR "This step was skipped, for the file 2.split_genome_seqs_and_get_cmds.ok exists.\n";
+}
+
+# 3 并行化运行 diamond blastx 命令
+# 按内存余量确定 diamond 的并行化数量。按单个 dianmond 命令峰值消耗 10 Gb 内存。
+print STDERR "3. run diamond blastx.\n";
+my $MemAvailable = &get_MemAvailable();
+my $paraFly_CPU = 1;
+$paraFly_CPU = $cpu / 4 if $paraFly_CPU < ($cpu / 4);
+$paraFly_CPU = $MemAvailable / 10000000 if $paraFly_CPU > ($MemAvailable / 10000000);
+$paraFly_CPU = 1 if $paraFly_CPU < 1;
+$paraFly_CPU = int($paraFly_CPU + 0.5);
+my $cmdString = "ParaFly -c split_genome_seqs/command.diamond_blastx.list -CPU $paraFly_CPU 2> 3.ParaFly_diamond_blastx.log";
+unless ( -e "1.3.ParaFly_diamond_blastx.ok" ) {
+    print STDERR (localtime) . ": CMD: $cmdString\n";
+    if ( system("$cmdString") != 0 ) {
+        $cmdString = "ParaFly -c split_genome_seqs/command.diamond_blastx.list -CPU 1 2>> 3.ParaFly_diamond_blastx.log";
+        print STDERR (localtime) . ": CMD: $cmdString\n";
+        system("$cmdString") == 0 or die "failed to execute: $cmdString\n";
+    }
+    open OUT, ">", "3.ParaFly_diamond_blastx.ok" or die $!; close OUT;
+}
+else {
+    print STDERR "CMD(Skipped): $cmdString\n";
+}
+
+# 4 汇总dimond的比对结果，并修正匹配的位置信息(之前打断基因组序列后，比对的位置需要进行修正)。
+print STDERR "4. output the alignment results.\n";
+foreach (<"split_genome_seqs/*.out">) {
+    open IN, $_ or die "Can not open file $_, $!";
+    while (<IN>) {
+        @_ = split /\t/;
+        my $pos = 0;
+        $pos = $1 if $_[0] =~ s/_start_(\d+)//;
+        $_[6] += $pos;
+        $_[7] += $pos;
+        print (join "\t", @_);
+    }
+    close IN;
+}
+
+sub get_MemAvailable {
+    open IN, "/proc/meminfo" or die "Can not open file /proc/meminfo, $!";
+    my $MemAvailable;
+    while (<IN>) {
+        if (m/MemAvailable:\s*(\d+)\s*kB/) {
+            $MemAvailable = $1;
+            next;
+        }
+    }
+    close IN;
+    return $MemAvailable;
+}
+
+sub get_partition {
+    my ($length, $ss, $os) = @_;
+    my @out;
+
+    my $pos = 1;
+    push @out, $pos;
+    while (($pos + $ss - 1) < $length) {
+        $pos = $pos + $ss - 1 - $os + 1;
+        push @out, $pos;
+    }
+    return @out;
+}