BioPHP - Oligonucleotide frequency based distance among sequences (original)
Original code submitted by josebaCode bellow is covered by GNU GPL v2 license.
Description
Last change: 2015/07/23 08:55 | Recent ChangesCompare sequences based in their oligonucleotide frecuencies (for 2-6 bases long oligos), get global distance (by computing a modified Pearson correlation), and apply UPGMA clustering to results. A table with distances and clustering results, including a dendrogram, are shown.
Code
Last change: 2015/07/23 08:55 | Download original | Recent Changes | Original code and
<!--
Title: Distance among sequences: compute global distance and show dendrogram
Author: Joseba Bikandi
License: GNU GLP v2
-->
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<title>Comparison of oligonucleotide composition in sequences</title>
</head>
<body bgcolor=FFFFFF>
<center>
<?php
// nothing has been posted -> show form and die
if (!$_POST){print_form(); die();}
// something has been posted -> go ahead
set_time_limit (10000);
error_reporting(0);
// to show computing period,
$timestart=date("U");
// GET SEQUENCES
// get data from form
$allsequences=$_POST["seq"];
// set a limit for maximum length of data posted
if (strlen($allsequences)>2000000){
die("Error: this service does not handle input requests longer than 2,000,000 bp.");
}
//remove a couple of things from sequence
$allsequences=substr($allsequences,strpos($allsequences,">")); // whatever is before ">", which is the start of the first sequence
$allsequences=preg_replace("/\r/","",$allsequences); // remove carriage returns ("\r"), but do not remove line feeds ("\n")
// split the individual sequences into $seqs array
$seqs=preg_split("/>/",$allsequences,-1,PREG_SPLIT_NO_EMPTY);
// get the name of each sequence (save names to array $seq_name)
foreach ($seqs as $key => $val){
$seq_name[$key]=substr($val,0,strpos($val,"\n"));
$temp_val=substr($val,strpos($val,"\n"));
$temp_val=preg_replace("/\W|\d/","",$temp_val);
$seqs[$key]=strtoupper($temp_val);
}
// at this moment two arrays are available: $seqs (with sequences) and $seq_names (with name of sequences)
// COMPUTE DISTANCES
// GET METHOD
if ($_POST["method"]=="euclidean"){ // EUCLIDEAN DISTANCE
// COMPUTE OLIGONUCLEOTIDE FREQUENCIES
foreach($seqs as $key => $val){
// to compute oligonucleotide frequencies, both strands are used
$seq_and_revseq=$val." ".RevComp($val);
$oligo_array[$key]= oligo_frequencies_standar($seq_and_revseq,$_POST["len"]);
}
// COMPUTE DISTANCES AMONG SEQUENCES
// by computing Euclidean distance
// standarized oligonucleotide frequencies in $oligo_array are used, and distances are stored in $data array
foreach($seqs as $key => $val){
foreach($seqs as $key2 => $val2){
if ($key>=$key2){continue;}
$data[$key][$key2]= Euclid_distance($oligo_array[$key],$oligo_array[$key2],$_POST["len"]);
}
}
}else{
// COMPUTE OLIGONUCLEOTIDE FREQUENCIES
foreach($seqs as $key => $theseq){
$oligo_array[$key]= compute_zscores_for_tetranucleotides($theseq);
}
// COMPUTE DISTANCES AMONG SEQUENCES
// by computing Pearson distance
// standarized oligonucleotide frequencies in $oligo_array are used, and distances are stored in $data array
foreach($seqs as $key => $val){
foreach($seqs as $key2 => $val2){
if ($key>=$key2){continue;}
$data[$key][$key2]= Pearson_distance($oligo_array[$key],$oligo_array[$key2]);
}
}
}
// DISPLAY TABLE WITH DISTANCES
print "<table border=1>\n<tr><td bgcolor=DDDDFF width=75>Distances</td>";
for($i=1;$i<sizeof($oligo_array);$i++){
print "<td bgcolor=DDDDFF width=75>$i</td>";
}
print "</tr>\n";
foreach($data as $key => $val){
print "<tr><td bgcolor=DDDDFF width=75>$key</td>";
for($i=1;$i<sizeof($oligo_array);$i++){
if ($data[$key][$i]){
$aa=floor(255*($data[$key][$i]));
print "<td style=\"background-color: rgb(255,$aa,$aa);\">".$data[$key][$i]."</td>";
}else{
print "<td> </td>";
}
}
print "</tr>\n";
}
print "</table>\n";
// DISPLAY NAME OF SEQUENCES
print "<table><tr><td>\n<p><b>Name of sequences</b>\n<pre>\n";
foreach($seq_name as $n => $name){
print " $n\t$name\n";
}
print "</pre></td></tr></table>\n";
// DISPLAY SELECTED OLIGONUCLEOTIDE LENGTH
print "Length of oligoncleotides for comparison: ".$_POST["len"];
// NEXT LINES WILL PERFORM UPGMA CLUSTERING
// in each loop, array $data is reduced (one case per loop)
while (sizeof($data)>1){
$min=min_array($data); // global variables are created: $x, $y, $min, $cases
$comp[$x][$y]=$min;
$data=new_array($data,$cases,$x,$y);
}
$min=min_array($data);
$comp[$x][$y]=$min;
// end of clustering
// array $comp stores the important data
// $textcluster is the results of the cluster as text.
// p.e.: ((3,4),7),(((5,6),1),2)
$textcluster="$x,$y";
print "<p>Clustering method: UPGMA<BR>$textcluster<p>";
// $max is the distance in the last clustering step (the root of the dendrogram)
$max=$a[$x][$y];
// CREATE THE IMAGE WITH THE DENDROGRAM
create_dendrogram ($textcluster,$comp,$max,$_POST["method"],$_POST["len"]);
// SHOW DENDROGRAM
print "<img src=image.png?".date("U")." border=1>";
// SHOW TIME REQUIRED FOR COMPUTING
$timetotal=date("U")-$timestart;
print "<p>Computed in $timetotal seconds<br>";
// #######################################################################################
// ############################## FUNCTIONS ####################################
// #######################################################################################
function get_cases($a){
$done="";
foreach($a as $key => $val){
$done.="#$key";
foreach($a[$key] as $key2 =>$val2){
$done.="#$key2";
}
}
$cases=preg_split("/#/",$done,-1,PREG_SPLIT_NO_EMPTY);
$cases=array_unique($cases);
sort($cases);
return $cases;
}
//#######################################################################################
function new_array($a,$cases,$x,$y){
$cases=get_cases($a);
for($j=0; $j<sizeof($cases)+1;$j++){
$key=$cases[$j];
// next 3 lines are required in windows for correct comparison
settype($key, "string");
settype($x, "string");
settype($y, "string");
if ($key==$x or $key==$y){continue;}
if ($a[$key][$x]!=""){
if ($a[$key][$y]!=""){$temp_a[$key]["($x,$y)"]=($a[$key][$x]+$a[$key][$y])/2;}
if ($a[$x][$key]!=""){$temp_a[$key]["($x,$y)"]=($a[$key][$x]+$a[$x][$key])/2;}
if ($a[$y][$key]!=""){$temp_a[$key]["($x,$y)"]=($a[$key][$x]+$a[$y][$key])/2;}
}else{
if ($a[$key][$y]!=""){
if ($a[$x][$key]!=""){$temp_a[$key]["($x,$y)"]=($a[$key][$y]+$a[$x][$key])/2;}
if ($a[$y][$key]!=""){$temp_a[$key]["($x,$y)"]=($a[$key][$y]+$a[$y][$key])/2;}
}else{
if ($a[$y][$key]!=""){$temp_a[$key]["($x,$y)"]=($a[$y][$key]+$a[$y][$key])/2;}
}
}
for($i=$j+1; $i<sizeof($cases);$i++){
$key2=$cases[$i];
settype($key2, "string");
if ($key==$key2 or $key2==$x or $key2==$y){continue;}
if ($a[$key][$key2]!=""){$temp_a[$key][$key2]=$a[$key][$key2];}
if ($a[$key2][$key]!=""){$temp_a[$key][$key2]=$a[$key2][$key];}
}
}
return $temp_a;
}
//#######################################################################################
function min_array($a){
global $x, $y, $min;
global $cases; // an array for cases
$str_cases="";
$min=1000000;
foreach($a as $key =>$val){
$str_cases.="#$key";
foreach($a[$key] as $key2 =>$val2){
if ($val==""){continue;}
$str_cases.="#$key2";
if ($val2<$min){
$min=$val2;
$x=$key;
$y=$key2;
}
}
}
$cases=preg_split("/#/",$done,-1,PREG_SPLIT_NO_EMPTY);
$cases=array_unique($cases);
sort($cases);
$min2=$min/2;
return $min;
}
//#######################################################################################
function create_dendrogram($str,$comp,$max,$method,$len){
$w=20; //height for each line (case)
$str=preg_replace("/\(|\)/","",$str).",";
$a=preg_split("/,/",$str,-1,PREG_SPLIT_NO_EMPTY);
$rows=sizeof($a);
$width=600; // width of scale from 0 to 2
$im = @imagecreatetruecolor($width*1.2, $rows*$w+40) or die("Unable to start image. It is GD available?");
$white =imagecolorallocate($im, 255, 255, 255);
$black =imagecolorallocate($im, 0, 0, 0);
$red =imagecolorallocate($im, 255, 0, 0);
imagefilledrectangle($im,0,0,$width*1.2, $rows*$w+40,$white);
$y=$rows*$w; // vertical location
$f=$width; // multiplication factor
// lines for scale
$j=0.1;
imageline ($im, log($j+1)*$f+20, $y, log($j+1)*$f+20, $y+10, $black);
imagestring($im, 1, log($j+1)*$f-8+20, $y+12, $j, $black);
$j=0.2;
imageline ($im, log($j+1)*$f+20, $y, log($j+1)*$f+20, $y+10, $black);
imagestring($im, 1, log($j+1)*$f-8+20, $y+12, $j, $black);
$j=0.3;
imageline ($im, log($j+1)*$f+20, $y, log($j+1)*$f+20, $y+10, $black);
imagestring($im, 1, log($j+1)*$f-8+20, $y+12, $j, $black);
$j=0.5;
imageline ($im, log($j+1)*$f+20, $y, log($j+1)*$f+20, $y+10, $black);
imagestring($im, 1, log($j+1)*$f-8+20, $y+12, $j, $black);
$j=1.0;
imageline ($im, log($j+1)*$f+20, $y, log($j+1)*$f+20, $y+10, $black);
imagestring($im, 1, log($j+1)*$f-8+20, $y+12, "1.0", $black);
$j=1.5;
imageline ($im, log($j+1)*$f+20, $y, log($j+1)*$f+20, $y+10, $black);
imagestring($im, 1, log($j+1)*$f-8+20, $y+12, $j, $black);
$j=2.0;
imageline ($im, log($j+1)*$f+20, $y, log($j+1)*$f+20, $y+10, $black);
imagestring($im, 1, log($j+1)*$f-8+20, $y+12, "2.0", $black);
// write into the image the numbers corresponding to cases
foreach($a as $n => $val){
if (strlen($val)==1){$val=" $val";}
imagestring($im,3, 5, $n*$w+5, $val, $black);
}
// WRITE LINES
foreach ($comp as $key => $val){
$pos1=$pos2=0;
foreach ($comp[$key] as $key2 => $val2){
// get position of case in the list
$keya=preg_replace("/\(|\)/","",$key);
$pos1=substr_count (" ,".substr ($str, 0,strpos(" ,".$str,",$keya,")),",")-0.4;
$keyb=preg_replace("/\(|\)/","",$key2);
$pos2=substr_count (" ,".substr ($str, 0,strpos(" ,".$str,",$keyb,")),",")-0.4;
if (substr_count($keya,",")>0){$pos1b=$pos1+substr_count($keya,",")/2;}else{$pos1b=$pos1;}
if (substr_count($keyb,",")>0){$pos2b=$pos2+substr_count($keyb,",")/2;}else{$pos2b=$pos2;}
// Position related data
$xkey1=$wherex[$key];
if ($xkey1==""){$xkey1=0;}
$xkey2=$wherex[$key2];
if ($xkey2==""){$xkey2=0;}
$max=max($xkey1,$xkey2);
$min=min($xkey1,$xkey2);
$xmax=$max+(($val2-($max))/2);
$val4=log($xmax+1)*$f;
$val4max=log($max+1)*$f;
$val4min=log($min+1)*$f;
// write lines
if ($wherex[$key]==$max){
imageline ($im, $val4max+20, $pos1b*$w, $val4+20, $pos1b*$w, $black); // ---
imageline ($im, $val4+20, $pos1b*$w, $val4+20, $pos2b*$w, $black); // |
imageline ($im, $val4min+20, $pos2b*$w, $val4+20, $pos2b*$w, $black); // ---
}else{
imageline ($im, $val4min+20, $pos1b*$w, $val4+20, $pos1b*$w, $black); // ---
imageline ($im, $val4+20, $pos1b*$w, $val4+20, $pos2b*$w, $black); // |
imageline ($im, $val4max+20, $pos2b*$w, $val4+20, $pos2b*$w, $black); // ---
}
$wherex["(".$key.",".$key2.")"]=$xmax;
}
}
imageline ($im, $val4+20, ($pos1b+$pos2b)*$w/2, $val4+40, ($pos1b+$pos2b)*$w/2, $black);
imageline ($im, 20, $y, $width*1.2, $y, $black);
if ($method=="euclidean"){
imagestring($im, 2, 5, $rows*$w+25, "Euclidean distance for $len bases long oligonucleotides.", $red);
}else{
imagestring($im, 2, 5, $rows*$w+25, "Pearson distance for z-scores of tetranucleotides.", $red);
}
imagestring($im, 2, $width*1, $rows*$w+25, "by insilico.ehu.es", black);
imagepng($im,"image.png");
imagedestroy($im);
}
//#######################################################################################
function RevComp($code){
$code=strrev($code);
$code=str_replace("A", "t", $code);
$code=str_replace("T", "a", $code);
$code=str_replace("G", "c", $code);
$code=str_replace("C", "g", $code);
$code = strtoupper ($code);
return $code;
}
//#######################################################################################
function Euclid_distance($a,$b,$k){
// Wang et al, Gene 2005; 346:173-185
$c=sqrt(pow(2,$k))/pow(4,$k); // contant
$sum=0;
foreach($a as $key => $val){
$sum+= pow($val-$b[$key],2);
}
return $c*sqrt($sum);
}
//#######################################################################################
function Pearson_distance($vals_x,$vals_y){
// normal correlation
if (sizeof($vals_x)!= sizeof($vals_y)){return;}
$sum_x=0;
$sum_x2=0;
$sum_y=0;
$sum_y2=0;
$sum_xy=0;
$n=sizeof($vals_x);
foreach($vals_x as $key => $val){
$val_x=$val;
$val_y=$vals_y[$key];
$sum_x+=$val_x;
$sum_x2+=$val_x*$val_x;
$sum_y+=$val_y;
$sum_y2+=$val_y*$val_y;
$sum_xy+=$val_x*$val_y;
//print "$val_x\t$val_y\n";
}
// calculate regression
$regresion=($sum_xy-(1/$n)*$sum_x*$sum_y)/((sqrt($sum_x2-(1/$n)*$sum_x*$sum_x)*(sqrt($sum_y2-(1/$n)*$sum_y*$sum_y))));
if ($regresion>0.999999999){$regresion=1;} // round data
return (1-$regresion);
}
//#######################################################################################
function compute_zscores_for_tetranucleotides($theseq){
// as described by Teeling et al. BMC Bioinformatics 2004, 5:163.
$theseq.=" ".RevComp($theseq);
$i=0;
$len=strlen($theseq)-2+1;
while ($i<$len){
$seq=substr($theseq,$i,2);
$oligos2[$seq]++;
$i++;
}
$i=0;
$len=strlen($theseq)-3+1;
while ($i<$len){
$seq=substr($theseq,$i,3);
$oligos3[$seq]++;
$i++;
}
$i=0;
$len=strlen($theseq)-4+1;
while ($i<$len){
$seq=substr($theseq,$i,4);
$oligos4[$seq]++;
$i++;
}
$base_a=array("A","C","G","T");
$base_b=array("A","C","G","T");
$base_c=array("A","C","G","T");
$base_d=array("A","C","G","T");
$base_e=array("A","C","G","T");
$base_f=array("A","C","G","T");
// COMPUTE Z-SCORES FOR TETRANUCLEOTIDES
$i=0;
foreach($base_a as $key_a => $val_a){
foreach($base_b as $key_b => $val_b){
foreach($base_c as $key_c => $val_c){
foreach($base_d as $key_d => $val_d){
$exp[$val_a.$val_b.$val_c.$val_d] = ($oligos3[$val_a.$val_b.$val_c]*$oligos3[$val_b.$val_c.$val_d])/$oligos2[$val_b.$val_c];
$var[$val_a.$val_b.$val_c.$val_d] = $exp[$val_a.$val_b.$val_c.$val_d] *((($oligos2[$val_b.$val_c]-$oligos3[$val_a.$val_b.$val_c])*($oligos2[$val_b.$val_c]-$oligos3[$val_b.$val_c.$val_d]))/pow($oligos2[$val_b.$val_c],2));
$zscore[$i] = ($oligos4[$val_a.$val_b.$val_c.$val_d]-$exp[$val_a.$val_b.$val_c.$val_d])/sqrt($var[$val_a.$val_b.$val_c.$val_d]);
$i++;
}}}}
return $zscore;
}
//#######################################################################################
function oligo_frequencies_standar($cadena,$len_oligos){
$i=0;
$len=strlen($cadena)-$len_oligos+1;
while ($i<$len){
$seq=substr($cadena,$i,$len_oligos);
$oligos_internos[$seq]++;
$i++;
}
$base_a=array("A","C","G","T");
$base_b=array("A","C","G","T");
$base_c=array("A","C","G","T");
$base_d=array("A","C","G","T");
$base_e=array("A","C","G","T");
$base_f=array("A","C","G","T");
//para oligos de 2
if ($len_oligos==2){
foreach($base_a as $key_a => $val_a){
foreach($base_b as $key_b => $val_b){
if ($oligos_internos[$val_a.$val_b]){
$oligos[$val_a.$val_b] = $oligos_internos[$val_a.$val_b];
}else{
$oligos[$val_a.$val_b] = 0;
}
}}
}
//para oligos de 3
if ($len_oligos==3){
foreach($base_a as $key_a => $val_a){
foreach($base_b as $key_b => $val_b){
foreach($base_c as $key_c => $val_c){
if ($oligos_internos[$val_a.$val_b.$val_c]){
$oligos[$val_a.$val_b.$val_c] = $oligos_internos[$val_a.$val_b.$val_c];
}else{
$oligos[$val_a.$val_b.$val_c] = 0;
}
}}}
}
//para oligos de 4
if ($len_oligos==4){
foreach($base_a as $key_a => $val_a){
foreach($base_b as $key_b => $val_b){
foreach($base_c as $key_c => $val_c){
foreach($base_d as $key_d => $val_d){
if ($oligos_internos[$val_a.$val_b.$val_c.$val_d]){
$oligos[$val_a.$val_b.$val_c.$val_d] = $oligos_internos[$val_a.$val_b.$val_c.$val_d];
}else{
$oligos[$val_a.$val_b.$val_c.$val_d] = 0;
}
}}}}
}
//para oligos de 5
if ($len_oligos==5){
foreach($base_a as $key_a => $val_a){
foreach($base_b as $key_b => $val_b){
foreach($base_c as $key_c => $val_c){
foreach($base_d as $key_d => $val_d){
foreach($base_e as $key_e => $val_e){
if ($oligos_internos[$val_a.$val_b.$val_c.$val_d.$val_e]){
$oligos[$val_a.$val_b.$val_c.$val_d.$val_e] = $oligos_internos[$val_a.$val_b.$val_c.$val_d.$val_e];
}else{
$oligos[$val_a.$val_b.$val_c.$val_d.$val_e] = 0;
}
}}}}}
}
//para oligos de 6
if ($len_oligos==6){
foreach($base_a as $key_a => $val_a){
foreach($base_b as $key_b => $val_b){
foreach($base_c as $key_c => $val_c){
foreach($base_d as $key_d => $val_d){
foreach($base_e as $key_e => $val_e){
foreach($base_f as $key_f => $val_f){
if ($oligos_internos[$val_a.$val_b.$val_c.$val_d.$val_e.$val_f]){
$oligos[$val_a.$val_b.$val_c.$val_d.$val_e.$val_f] = $oligos_internos[$val_a.$val_b.$val_c.$val_d.$val_e.$val_f];
}else{
$oligos[$val_a.$val_b.$val_c.$val_d.$val_e.$val_f] = 0;
}
}}}}}}
}
$oligos=standar_frecuencies($oligos, $len_oligos);
return $oligos;
}
function standar_frecuencies($array, $m){
$sum=0;
foreach($array as $k => $v){
$sum+=$v;
}
$c=pow(4,$m)/$sum;
foreach($array as $k => $v){
$array[$k]= $c*$v;
}
return $array;
}
// #######################################################################################
function print_form(){
?>
<table cellpadding=10>
<tr><td bgcolor=ddddff>
<h1>Distance between sequences:<br>comparison of oligonucleotide composition</h1>
<ul>
<li>This tool will compute distance between input sequences and UPGMA clustering will be applied.</li>
<li>Distances are shown in a table, and a dendrogram is displaied.</li>
<li>Check carefully the results. Theirvalue for phylogenetics is limited.</li>
</ul>
<center>
<form method="post" action="<? print $_SERVER["PHP_SELF"]; ?>">
Copy your sequences bellow as Fasta (Maximum length: 2 MB)<br>
<textarea name="seq" cols="80" rows="10"></textarea>
</center>
<p>Choose method to compute distances (<a href=http://insilico.ehu.es/oligoweb/info/distance.php>info</a>)
<br>
<input type=radio name=method value=pearson> Pearson distance for z-scores of tetranucleotides (>20000 bp sequences)
<br>
<input type=radio name=method value=euclidean checked>Euclidean distance for
<select name=len>
<option value=2>dinucleotides
<option value=3>trinucleotides
<option value=4 selected>tetranucleotides
<option value=5>pentanucleotides
<option value=6>hexanucleotides
</select><p>
<center>
<p><input value="Compare sequences" type="submit">
</form>
</center>
<div align=right><a href=http://insilico.ehu.es/oligoweb/my_distance/example.php>Example</a></div>
</td></tr><tr><td>
This is a simplified PHP script extracted from the online tool at <a href=http://insilico.ehu.es/oligoweb/>insilico.ehu.es/oligoweb/</a>.
<br>Compare your sequence to up-to-date sequenced prokaryotes <a href=http://insilico.ehu.es/oligoweb/my_distance/>here</a>.
<p>Source code is available at <a href=http://www.biophp.org/minitools/distance_among_sequences/>BioPHP.org</a>
</td></tr></table>
</center>
</body>
</html>
<?
}
?>
</body>
</html>