Tryag File Manager

/home/cluster1/data/bu01/1121861/jlex/rdloci/php5/repeat_analyzer.php5

class repeat_analyzer {
    public $MAX_MISMATCHES;
    public $FONT_SIZE;
    public $colors = array("pink","green","blue","red","cyan","gray","purple","gold","brown","#FF8C00", "#8B0000");
    public $light_colors = array("red"=>"salmon","cyan"=>"#E0FFFF", "green"=>"#ADFF2F", "blue"=>"#4682B4","gray"=>"silver","pink"=>"#FFE4E1", "purple"=>"#EE82EE", "gold"=>"yellow","brown"=>"tan", "#FF8C00"=>"orange", "#8B0000"=>"#B22222");

$out = fopen("variable_regions.txt","w");
	foreach($seqs as $name=>$seq) {
	    $seq = ereg_replace(" ","",$seq);
	    $seq = ereg_replace("\*","",$seq);
	    fwrite($out,">".$name."\n".$seq."\n\n");
	}
	fclose($out);
	$seqs["maps"] = $maps;
	return $seqs;
    }

$length = strlen($repeat);
		    $repeat_a = $this->str_split($repeat);
		    $position = 0;
		    foreach($repeat_a as $letter) {
			if(@!array_key_exists($letter,$by_repeat[$index][$position])) {
			    $by_repeat[$index][$position][$letter] = 1;
			}
			else {
			    $by_repeat[$index][$position][$letter]++;
			}
			$position++;
		    }
		}
	    }
	}
	foreach($by_repeat as $repeat=>$set) {
	    if(count($set) == 0) {
		unset($by_repeat[$repeat]);
	    }
	}
	return $by_repeat;
    }

/**
     * Counts the number of types of repeats in a set of sequences and returns
     * a sorted associated array, (repeat=>count), of highest count first. 
     */
    function count_repeats($seqs) {	
	$total_words = 0;

$repeats = array();
	foreach($seqs as $name=>$seq) {
	    $a = explode(" ",$seq);
	    foreach($a as $repeat) {
		$repeat = trim($repeat);
		if(($repeat != "") && (!ereg("[*]",$repeat))) {
		    $total_words++;
		    if(array_key_exists($repeat,$repeats)) {
			$repeats[$repeat]++;
		    }
		    else {
			$repeats[$repeat] = 1;
		    }
		}
	    }
	}
	arsort($repeats);
	return $repeats;
    }

function assign_colors($standards_file) {
	$colors = $this->colors;
	$light_colors = $this->light_colors;
	$standards = $this->load_seqs($standards_file);
	$repeats = $this->count_repeats($standards);

$reps = array();
	$sim_reps = array();
	$i = 0;
	foreach($repeats as $repeat=>$count) {
	    $repeat = trim($repeat);
	    if(!ereg("\*",$repeat) && ($repeat != "")) {
		$cur_color = "";
		foreach($reps as $rep=>$color) {
		    $num_mismatches = $this->count_mismatches($rep,$repeat);
		    if($num_mismatches <= $this->MAX_MISMATCHES) {
			$cur_color = $color;
			break;
		    }
		}
		if(($cur_color == "") && ($count > 1)) {
		    $reps[$repeat] = $colors[$i];
		    $i++;
		}
		else {
		    $sim_reps[$repeat] = $light_colors[$cur_color];
		}
	    }
	}
	$repeat_colors = array_merge($reps,$sim_reps);
	return $repeat_colors;
    }

function display_repeat_counts($repeat_counts, $repeat_colors) {
	$total_words = 0;
	foreach($repeat_counts as $repeat=>$count) {
	    $total_words += $count;
	}
	echo "<table border='1'><tr><td><b>word</b></td><td><b>count</b></td><td><b>% of total</b></td></tr>";	
	foreach($repeat_counts as $repeat=>$count) {
	    $percent = round((100*$count/$total_words),2);
	    echo "<tr><td bgcolor='".$repeat_colors[$repeat]."' NOWRAP='1'>$repeat</td><td>$count</td><td>$percent%</td></tr>";
	}
	echo "</table>";

$count_by_color = array();
	foreach($repeat_counts as $repeat=>$count) {
	    $color = $repeat_colors[$repeat];
	    $count_by_color[$color] += $count;
	}
	echo "<table border='1'>";

foreach($this->colors as $color) {
	    $light_color = $this->light_colors[$color];
	    $total = $count_by_color[$color] + $count_by_color[$light_color];
	    $percent = round((100*$total/$total_words),2);
	    if($total != 0) {
		echo "<tr><td bgcolor='$color' width='50'></td><td>+</td><td bgcolor='$light_color' width='50'></td><td>$total</td><td>$percent%</td></tr>";
	    }
	}
	echo "</table>";
    }

function display_repeats($seqs, $repeat_colors) {
	$font_size = $this->FONT_SIZE;
	echo "<table>";
	foreach($seqs as $name=>$seq) {
	    $a = explode(" ",$seq);
	    $strain = substr($name,0,strpos($name,"_"));
	    $name = ereg_replace(" ","_",$name);
	    $name = substr($name,0,-1);

echo "<tr><td><font style=\"font-size: $font_size;\" NOWRAP='1'><a href='../mysql_to_xml.php4?num_conditions=1&database=strains&phrases_and_fields_0=name&reg_exps_0=equals&string_0=$strain&stylesheet=strains/to_html.xml'>$name</a></b></font></td>";
	    foreach($a as $cur_repeat) {
		if(trim($cur_repeat) != "") {
		    if(ereg("[*]",$cur_repeat)) {
			echo "<td bgcolor='white'></td>";
		    }
		    else {
			$color = $repeat_colors[$cur_repeat];
			if(trim($color) == "") {
			    $color = "DDDDDD";
			}
			echo "<td bgcolor=\"$color\"><font style=\"font-size: $font_size;\">";
			$length = 2*strlen($cur_repeat);
			if($color == "DDDDDD") {
			    echo "-";
			}
			else {
			    for($q=0;$q<15;$q++) {
				echo "&nbsp;";
			    }
			}
			echo "</font></td>";
			//echo "$cur_repeat</font></td>";
		    }
		}
	    }
	    echo "</tr>";
	    echo "<tr></tr>";
	}
	echo "</table>";
    }

function analyze($seqs_file, $standards_file, $font_size) {
	$repeat_colors = $this->assign_colors($standards_file);
	
	$seqs = $this->load_seqs($seqs_file);
	$maps = $seqs["maps"];
	unset($seqs["maps"]);
	$repeat_counts = $this->count_repeats($seqs);
	$repeat_patterns = $this->find_patterns($seqs,$repeat_colors);
	echo "<table>";
	echo "<tr><td>";
	$this->display_locus_maps($maps);
	echo"</td></tr>";
	echo "<tr><td colspan='2'>";
	$this->display_repeats($seqs, $repeat_colors);
	echo "</td></tr>";
	echo "<tr><td valign='top'>";
	$this->display_repeat_counts($repeat_counts, $repeat_colors);
	echo "</td><td>";
	$this->display_patterns($repeat_patterns,$repeat_colors);
	echo "</td></tr></table>";
     }
}
/*
error_reporting(0);
$file = $_POST["file"];
$standards = $_POST["standards"];
echo $stardards."<BR>";
$size = $_POST["size"];
$max_mismatches = $_POST["max_mismatches"];
$ca = new repeat_analyzer($max_mismatches, $size);
$ca->analyze("../vregions/$file","../vregions/$standards",$size);
*/