1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713
|
#!/usr/bin/env perl
# $Id: evaluate_model,v 1.15 2006/12/19 13:15:06 c4chris Exp $
################################################################################
#
# evaluate_model
# --------------
#
# Claudio Lottaz, SIB-ISREC, Claudio.Lottaz@isb-sib.ch
# Christian Iseli, LICR ITO, Christian.Iseli@licr.org
#
# Copyright (c) 1999, 2006 Swiss Institute of Bioinformatics.
# All rights reserved.
#
################################################################################
use strict;
use FASTAFile;
use Symbol;
# global configuration
my $winsegshuffle_dbsize = 10000000; # the size for shuffled databases
# global variables
my $verbose = 1;
my $norna = 0;
my $optionsfile = '';
my $forcedtuplesize = undef;
my $forcedminmask = undef;
my $forcedpseudocounts = undef;
my $forcedminscore = undef;
my $forcedstartlength = undef;
my $forcedstartpreroll = undef;
my $forcedstoplength = undef;
my $forcedstoppreroll = undef;
my $forcedsmatfile = undef;
my $forceestscanparams = undef;
my $datadir = '.';
my $filestem = '';
my $program = "estscan";
require "build_model_utils.pl";
################################################################################
#
# Check command-line for switches
#
my $usage = "Usage: evaluate_model [options] <conf-files>\n" .
" where options are:\n" .
" -q don't log on terminal\n" .
" -O <str> options passed to program\n" .
" -o <file> options file\n" .
" -n skip evaluation on mRNAs\n" .
" -t <int> force tuple size, overwrites entry in config-files\n" .
" -M <file> force score matrices file\n" .
" -m <int> force minimal mask, overwrites entry in config-files\n" .
" -P <file> force program name (or file)\n" .
" -p <int> force pseudocounts, overwrites entry in config-files\n" .
" -s <int> force minimal score, overwrites entry in config-files\n" .
" -l <int> force length of start profile (in codons/triplets)\n" .
" -r <int> force start profile's preroll in 5'UTR (in codons/triplets)\n" .
" -L <int> force length of stop profile (in codons/triplets)\n" .
" -R <int> force sop profile's preroll in 5'UTR (in codons/triplets)\n" .
"More information is obtained using 'perldoc evaluate_model'\n";
while ($ARGV[0] =~ m/^-/) {
if ($ARGV[0] eq '-q') { shift; $verbose = 0; next; }
if ($ARGV[0] eq '-n') { shift; $norna = 1; next; }
if ($ARGV[0] eq '-O') { shift; $forceestscanparams = shift; next; }
if ($ARGV[0] eq '-o') { shift; $optionsfile = shift; next; }
if ($ARGV[0] eq '-t') { shift; $forcedtuplesize = shift; next; }
if ($ARGV[0] eq '-M') { shift; $forcedsmatfile = shift; next; }
if ($ARGV[0] eq '-m') { shift; $forcedminmask = shift; next; }
if ($ARGV[0] eq '-P') { shift; $program = shift; next; }
if ($ARGV[0] eq '-p') { shift; $forcedpseudocounts = shift; next; }
if ($ARGV[0] eq '-s') { shift; $forcedminscore = shift; next; }
if ($ARGV[0] eq '-l') { shift; $forcedstartlength = shift; next; }
if ($ARGV[0] eq '-r') { shift; $forcedstartpreroll = shift; next; }
if ($ARGV[0] eq '-L') { shift; $forcedstoplength = shift; next; }
if ($ARGV[0] eq '-R') { shift; $forcedstoppreroll = shift; next; }
die "Unrecognized switch: $ARGV[0]\n$usage";
}
if ($#ARGV < 0) { die "No configuration file specified\n$usage"; }
################################################################################
#
# Main-loop through all specified config-files
#
my $parFile;
while($parFile = shift) {
my($organism, $hightaxo, $dbfiles, $ugdata, $estdata, $datadir2, $filestem2,
$rnafile, $estfile, $estcdsfile, $estutrfile, $trainingfile, $testfile,
$utrfile, $cdsfile, $tuplesize, $minmask, $pseudocounts, $minscore,
$startlength, $startpreroll, $stoplength, $stoppreroll, $smatfile,
$estscanparams, $nb_isochores, $isochore_borders) =
readConfig($parFile, $forcedtuplesize, $forcedminmask,
$forcedpseudocounts, $forcedminscore,
$forcedstartlength, $forcedstartpreroll,
$forcedstoplength, $forcedstoppreroll, $verbose);
if (defined $forcedsmatfile) {
$smatfile = $forcedsmatfile;
}
if (defined $forceestscanparams) {
$estscanparams = $forceestscanparams;
}
log_open("readconfig.log");
$datadir = $datadir2; $filestem = $filestem2;
showConfig($parFile, $organism, $hightaxo, $dbfiles, $ugdata, $estdata, $datadir,
$rnafile, $estfile, $estcdsfile, $estutrfile, $trainingfile,
$testfile, $utrfile, $cdsfile, $tuplesize, $minmask,
$pseudocounts, $minscore, $startlength, $startpreroll,
$stoplength, $stoppreroll, $smatfile, $estscanparams,
$nb_isochores, $isochore_borders);
if (system("which $program") != 0) {
log_print(" fatal: $program not found");
die;
}
log_close();
# read options-file, each line contains a set of options
my $o; my @options;
if ($optionsfile eq "") {
push @options, $estscanparams;
} else {
my $fh = gensym;
open $fh, $optionsfile;
@options = <$fh>;
close $fh;
}
# evaluate
log_open("evaluate_model.log");
log_print("\nUsing $program to scan EST/mRNA\n");
my %uf = ( 'rna' => $utrfile, 'est' => $estutrfile );
my %cf = ( 'rna' => $cdsfile, 'est' => $estcdsfile );
my %tf = ( 'rna' => $testfile, 'est' => $estcdsfile );
my %N = ( 'rna' => "mRNA", 'est' => "EST" );
foreach my $k ("rna", "est") {
next if ($k eq "rna") && ($norna == 1);
next unless -s $cf{$k};
foreach $o (@options) {
$o =~ s/\n$//;
log_print("\nEvaluating new model on "
. $N{$k} . " data using $o....");
estimateFalse($k, 'utr', $uf{$k}, $smatfile, $o);
estimateFalse($k, 'cds', $cf{$k}, $smatfile, $o);
evaluateStartStop($smatfile, $tf{$k}, $k, $o);
log_print("\nEvaluating new model on "
. $N{$k} . " data using $o -S....");
estimateFalse($k, 'utr', $uf{$k}, $smatfile, "$o -S ");
estimateFalse($k, 'cds', $cf{$k}, $smatfile, "$o -S ");
evaluateStartStop($smatfile, $tf{$k}, $k, "$o -S ");
if ($#options > 0) {
unlink <$datadir/Evaluate/${k}prc*>;
}
}
}
log_close();
print "$parFile done.\n";
}
exit 0;
sub estimateFalse {
# Runs ESTScan on ESTs from UTRs and CDSs. According to $cdsutr
# determines the false positive or false negative rate.
my ($prefix, $cdsutr, $estfile, $smatfile, $estscanparams) = @_;
my $x = `grep -i '^[a-z]' $estfile | wc`;
$x =~ s/^\s+//;
my($l, $w, $c) = split(/\s+/, $x);
my $nbTot = $c - $l;
# compute predictions for estfile
my $EP = $estscanparams;
$EP =~ s/[- ]+//g;
my $prcfile = "$datadir/Evaluate/$prefix"
. "prc$filestem$EP" . "_$cdsutr.seq";
if (-s $prcfile) {
log_print(" - $prcfile already exists, skipped");
} else {
log_print(" - predicting CDS for $estfile...");
system "$program $estscanparams -M $smatfile $estfile > $prcfile";
}
$x = `grep -i '^[a-z]' $prcfile | wc`;
$x =~ s/^\s+//;
($l, $w, $c) = split /\s+/, $x;
my $nbPrc = $c - $l;
my $r = $nbPrc/$nbTot;
my $nbSeq = `grep -c '^>' $prcfile`;
chop $nbSeq;
my $nbTotSeq = `grep -c '^>' $estfile`;
chop $nbTotSeq;
log_print(" found $nbPrc coding of $nbTot nucleotides " .
"in $nbSeq of $nbTotSeq sequences");
if ($cdsutr eq 'utr') {
my $pct = sprintf "%.2f", 100 * $r;
my $pctseq = sprintf "%.2f", 100 * ($nbSeq / $nbTotSeq);
log_print(" estimated false positive rate: $pct% (nt) $pctseq% (seq)");
} else {
my $pct = sprintf "%.2f", 100 * (1.0 - $r);
my $pctseq = sprintf "%.2f", 100 * (1.0 - ($nbSeq / $nbTotSeq));
log_print(" estimated false negative rate: "
. "$pct% (nt) $pctseq% (seq)");
}
# read scores from prcfile
my @scores;
my $fh = gensym;
open $fh, $prcfile;
while ( <$fh> ) {
if (m/^>\S+ (\S+) (\S+) (\S+)/) {
push @scores, $1;
}
# if (m/^>\S+ (\S+) (\S+) (\S+)/) {
# push @scores, $1/($3 - $2 + 1);
# } # length norm
}
close $fh;
# write cumulative data
my $score;
my $cumulator = ($cdsutr eq 'utr') ? $#scores + 1 : 0;
my $histofile = "$datadir/Evaluate/$prefix" . "prc$filestem";
if ($cdsutr eq 'utr') {
open $fh, ">$histofile.dat";
} else {
open $fh, ">>$histofile.dat";
print $fh "\n";
}
foreach $score (sort {$a <=> $b} @scores) {
my $y = 100 * $cumulator / ($#scores + 1);
if (($y < 90) && ($y > 1)) {
print $fh "$score $y\n";
}
if ($cdsutr eq 'utr') {
$cumulator--;
} else {
$cumulator++;
}
}
close $fh;
# write gnuplot script
my($order, $minmask, $pseudo) = split /\_/, $filestem;
my $title = "Selectivity on UTRs / Sensitivity on CDSs "
. "(order $order, minmask $minmask, pseudocounts $pseudo)";
open $fh, ">$histofile.gplot";
print $fh <<EndOfBuffer;
set title "$title".
set xlabel "Score"
set ylabel "Percentage"
set data style lines
set nokey
plot '${prefix}prc${filestem}.dat'
EndOfBuffer
close $fh;
}
################################################################################
#
# Estimate false negative rate on RNA data
# Evaluate the the model's accuracy on start/stop detection in RNA data
#
sub evaluateStartStop {
# on RNA data test how close ESTScan detects the
# beginning and the end of the coding region
my ($smatfile, $testfile, $prefix, $estscanparams) = @_;
# predict on RNA/EST data
my $EP = $estscanparams;
$EP =~ s/[- ]+//g;
my $resultfile = "$datadir/Evaluate/$prefix" . "prc$filestem$EP.seq";
$resultfile =~ s/ //g;
if (-s $resultfile) {
log_print(" - $resultfile already exists, skipped");
} else {
log_print(" - predicting CDS on $testfile...");
system "$program $estscanparams -M $smatfile $testfile > $resultfile";
}
# count in testfile
my(%seqLengths, %seqCDSlen, $currSeq, $e);
my $origCtr = 0;
my $src = FASTAFile->new($testfile);
$src->openStream;
while (defined( $e = $src->getNext )) {
$seqLengths{$e->ac} = length $e->{_seq};
$e->{_seqHead} =~ m/CDS: (\S+) (\S+)/;
$seqCDSlen{$e->ac} = $2 - $1 + 1;
$origCtr++;
}
close $src->{_BTFfile};
if ($origCtr == 0) {
log_print(" no sequences found in $testfile, skipped");
return;
}
# initialize variables
my(%startHistogram, %stopHistogram);
my $delta = 99; # distances larger than this are considered weak predictions
my $averageStart = 0;
my $averageStop = 0;
my $weakStartCtr = 0;
my $wrongFrameCtr = 0;
my $weakStopCtr = 0;
my $worstSeq = "";
my $ctr = 0;
my $cdsNts = 0;
my $utrNts = 0;
my $fpNts = 0;
my $fnNts = 0;
# count utr for false positive/negative rate
if ($prefix eq 'est') {
my $utrfile = "$datadir/Evaluate/estutr.seq";
my $x = `grep -i '^[a-z]' $utrfile | wc`;
$x =~ s/^\s+//;
my($l, $w, $c) = split /\s+/, $x;
$utrNts = $c - $l;
$utrfile = "$datadir/Evaluate/$prefix" . "prc$filestem$EP" . "_utr.seq";
$x = `grep -i '^[a-z]' $utrfile | wc`;
$x =~ s/^\s+//;
($l, $w, $c) = split /\s+/, $x;
$fpNts = $c - $l;
}
# read and evaluate the result file
log_print(" - computing histograms from $resultfile...");
$src = FASTAFile->new($resultfile);
$src->openStream;
while (defined( $e = $src->getNext)) {
# read entry
#if ($e->{_FASTAde} =~ m/minus strand/) { next }
my ($predStart, $predStop, $annotatedStart, $annotatedStop) =
$e->{_seqHead} =~ m/(\S+) (\S+) ?CDS: (\S+) (\S+)/;
my $theId = $e->ac;
if ($e->{_FASTAde} =~ m/minus strand/ && $predStart ne "") {
my $tmp = $seqLengths{$e->ac} - $predStart + 1;
$predStart = $seqLengths{$e->ac} - $predStop + 1;
$predStop = $tmp;
}
# update histogram
my $diffStart = $predStart - $annotatedStart;
my $diffStop = $predStop - $annotatedStop;
$averageStart += abs($diffStart);
$averageStop += abs($diffStop);
if (abs($diffStart) > $delta) {
$weakStartCtr++;
}
if (abs($diffStop) > $delta) {
$weakStopCtr++;
}
if ((($diffStart % 3) != 0) && ($annotatedStart != 1)) {
$wrongFrameCtr++;
}
$startHistogram{sprintf("%05d", $diffStart)} += 1;
$stopHistogram{sprintf("%05d", $diffStop)} += 1;
$ctr++;
# update false positive/negative data
my $newCds = $annotatedStop - $annotatedStart + 1;
$cdsNts += $newCds;
$utrNts += $seqLengths{$e->ac} - $newCds;
if ($predStart ne ""
&& (($annotatedStart > $predStop)
|| ($annotatedStop < $predStart))) {
$fpNts += $predStop - $predStart + 1;
$fnNts += $annotatedStop - $annotatedStart + 1;
} else {
if ($diffStart < 0) {
$fpNts -= $diffStart;
} else {
$fnNts += $diffStart;
}
if ($diffStop < 0) {
$fnNts -= $diffStop;
} else {
$fpNts += $diffStop;
}
}
delete $seqCDSlen{$e->ac};
}
close $src->{_BTFfile};
foreach (values(%seqCDSlen)) {
$fnNts += $_;
}
log_print(" predicted $ctr coding regions");
my $pct = sprintf "%.2f", 100 * ($fpNts / $utrNts);
log_print(" estimated false positive rate $pct% (nt)");
$pct = sprintf "%.2f", 100 * ($fnNts / $cdsNts);
log_print(" estimated false negative rate $pct% (nt)");
$averageStart /= $ctr;
$averageStop /= $ctr;
log_print(" - writing data-files and gnuplot scripts...");
# write data for start histogram
my $fh = gensym;
my $histofile
= "$datadir/Evaluate/$prefix" . "prc$filestem" . "_starthisto";
open $fh, ">$histofile.dat";
foreach (sort (keys %startHistogram)) {
print $fh "$_ $startHistogram{$_}\n";
}
close $fh;
my $exactStart = $startHistogram{'00000'};
# write gnuplot script for start histogram
$averageStart = sprintf("%.f", $averageStart);
open $fh, ">$histofile.gplot";
print $fh <<EndOfBuffer;
set title "Start accuracy: analyzed $origCtr".
set label "Predicted $ctr coding sequences" at graph 0.95, 0.95 right
set label "Distance More than $delta in $weakStartCtr cases" at graph 0.95, 0.9 right
set label "Exact prediction in $exactStart cases" at graph 0.95, 0.85 right
set label "Wrong frame in $wrongFrameCtr cases" at graph 0.95, 0.8 right
set label "Average distance $averageStart" at graph 0.95, 0.75 right
set xlabel "Relative Position"
set ylabel "Frequency"
set data style boxes
set boxwidth 1
set nokey
plot [-$delta:$delta] '${prefix}prc${filestem}_starthisto.dat'
EndOfBuffer
close $fh;
# write data file for percentile picture of start histogram
$histofile .= '_p';
foreach (keys %startHistogram) {
if ($_ < 0) {
$startHistogram{sprintf("%05d", -$_)}
+= $startHistogram{sprintf("%05d", $_)};
delete $startHistogram{$_};
}
}
my $cumulus = $origCtr;
my $pieLabels = "c(";
my $pieValues = "c(";
my $x = 0;
my $y = 0;
my $distBuffer = '';
my @ticks = (10, 25, 50, 100, 100);
open $fh, ">$histofile.dat";
foreach (sort (keys %startHistogram)) {
$x += $startHistogram{$_};
$cumulus -= $startHistogram{$_};
print $fh "$_ ", $cumulus/$origCtr*100, "\n";
if (($_ >= $y) && ($#ticks > -1)) {
$distBuffer .= sprintf "\t<=%4d: %8d\n", $y, $x;
$pieLabels .= "'<=$y:" . sprintf("%4.1f\%", $x/$origCtr*100) . "', ";
$pieValues .= "$x, ";
$x = 0;
$y = shift @ticks;
}
}
close $fh;
$pieLabels .= "'>$y: " . sprintf("%4.1f\%", $x/$origCtr*100)
. "', 'missed: " . sprintf("%4.1f4\%", 100*$cumulus/$origCtr) . "')";
$pieLabels =~ s/<=0:/exact:/;
$pieValues .= "$x, $cumulus)";
$distBuffer .= sprintf "\t >%4d: %8d\n\tmissed: %8d", $y, $x, $cumulus;
log_print(" start predicted at distance\n$distBuffer");
# write R script for pie chart
my $piescript = "$datadir/Evaluate/$filestem\_piecharts.R";
if ($prefix eq 'rna') {
unlink $piescript;
}
open $fh, ">>$piescript";
print $fh <<EndOfBuffer;
postscript('${prefix}prc${filestem}_startpie.eps', horizontal=F,
onefile=F, width=3.0, height=3.0, paper="special", pointsize=8)
x <- $pieValues
pie(x, labels=$pieLabels, col=grey(seq(1,0,-1/length(x))))
dev.off()
EndOfBuffer
close $fh;
# write gnuplot script for percentile picture of start histogram
open $fh, ">$histofile.gplot";
print $fh <<EndOfBuffer;
set title "Start accuracy: Percentage of Predictions far away from Annotation".
set xlabel "Distance"
set ylabel "Percentage"
set data style lines
set nokey
plot [0:200] '${prefix}prc${filestem}_starthisto_p.dat'
EndOfBuffer
close $fh;
# write data file for stop histogram
$histofile = "$datadir/Evaluate/$prefix" . "prc$filestem" . "_stophisto";
open $fh, ">$histofile.dat";
foreach (sort (keys %stopHistogram)) {
print $fh "$_ $stopHistogram{$_}\n";
}
close $fh;
$averageStop = sprintf "%.1f", $averageStop;
my $exactStop = $stopHistogram{'00000'};
# write gnuplot script for stop histogram
open $fh, ">$histofile.gplot";
print $fh <<EndOfBuffer;
set title "Stop accuracy: analyzed $origCtr sequences".
set label "Predicted $ctr coding sequences" at graph 0.95, 0.95 right
set label "Distance More than $delta in $weakStopCtr cases" at graph 0.95, 0.9 right
set label "Exact prediction in $exactStop cases" at graph 0.95, 0.85 right
set label "Average distance $averageStop" at graph 0.95, 0.8 right
set xlabel "Relative Position"
set ylabel "Frequency"
set data style boxes
set boxwidth 1
set nokey
plot [-$delta:$delta] '${prefix}prc${filestem}_stophisto.dat'
EndOfBuffer
close $fh;
# write data file for percentage picture of stop histogram
$histofile .= '_p';
foreach (keys %stopHistogram) {
if ($_ < 0) {
$stopHistogram{sprintf("%05d", -$_)}
+= $stopHistogram{sprintf("%05d", $_)};
delete $stopHistogram{$_};
}
}
$x = 0;
$y = 0;
$distBuffer = '';
@ticks = (10, 25, 50, 100, 100);
$cumulus = $origCtr;
$pieLabels = "c(";
$pieValues = "c(";
open $fh, ">$histofile.dat";
foreach (sort (keys %stopHistogram)) {
$x += $stopHistogram{$_};
$cumulus -= $stopHistogram{$_};
print $fh "$_ ", $cumulus/$origCtr, "\n";
if (($_ >= $y) && ($#ticks > -1)) {
$distBuffer .= sprintf "\t<=%4d: %8d\n", $y, $x;
$pieLabels .= "'<=$y:" . sprintf("%4.1f\%", 100*$x/$origCtr) . "', ";
$pieValues .= "$x, ";
$x = 0;
$y = shift @ticks;
}
}
close $fh;
$pieLabels .= "'>$y: " . sprintf("%4.1f\%", $x/$origCtr*100)
. "', 'missed: " . sprintf("%4.1f4\%", 100*$cumulus/$origCtr) . "')";
$pieLabels =~ s/<=0:/exact:/;
$pieValues .= "$x, $cumulus)";
$distBuffer .= sprintf "\t >%4d: %8d\n\tmissed: %8d", $y, $x, $cumulus;
log_print(" stop predicted at distance\n$distBuffer");
# write R script for pie chart
open $fh, ">>$piescript";
print $fh <<EndOfBuffer;
postscript('${prefix}prc${filestem}_stoppie.eps', horizontal=F, onefile=F,
width=3.0, height=3.0, paper="special", pointsize=8)
x <- $pieValues
pie(x, labels=$pieLabels, col=grey(seq(1,0,-1/length(x))))
dev.off()
EndOfBuffer
close $fh;
# write gnuplot script for percentile picture of stop histogram
open $fh, ">$histofile.gplot";
print $fh <<EndOfBuffer;
set title "Stop accuracy: Percentage of Predictions far away from Annotation".
set xlabel "Distance"
set ylabel "Percentage"
set data style lines
set nokey
plot [0:200] '${prefix}prc${filestem}_stophisto_p.dat'
EndOfBuffer
close $fh;
}
################################################################################
#
# Documentation
#
=head1 NAME
evaluate_model - evaluate an ESTScan model generated by build_model
=head1 SYNOPSIS
evaluate_model [options] <config-files...>
=head1 DESCRIPTION
evaluate_model evaluates the performance of an ESTscan model. It uses
the same directory structure as build_model, knows most of
build_model's command line switches and read the same configuration
files. ESTScan's performance is evluated in terms of false positive
and negative rates on the nucleotide level, as well as prediction
accuracy for start and stop sites. The script reads configuration
files on the comannd line and performs the following steps for each of
them:
- Extract untranslated regions from test mRNA
- Evaluate false positive rate. evaluate false negative
- Evaluate false negative rate as well as start and stop prediction
accuracy on test mRNAs
- Find entirely untranslated and partially coding ESTs using UniGene
- Evaluate false positive rate on untranslated ESTs
- Evaluate false negative rate as well as start and stop prediction
accuracy on partially coding ESTs
The evaluation is carried out on two kinds of data: mRNA data and EST
data. The first three steps work on mRNA the latter on ESTs. When
using RNA, two computational experiments are made. On one hand all
untranslate regions of the test set of mRNAs are extracted and then
analyzed in order to find the false positive rate on nucleotide as
well as on sequence level. On the other hand, the complete test mRNAs
are analyzed to estimate false positive and false negative rates on a
nucleotide level. Moreover, distance distributions between prediction
and annotation of start and stop sites are computed and histograms
generated as gnuplot scripts and data files.
The data needed for evluation using ESTs is extracted using UniGene.
UniGene clusters are used to determine ESTs from untranslated regions
and coding sequence respectively. This is done by matching the ESTs of
a given cluster against its full-length mRNA with megablast and then
determining where the match occurs relative to the annotated coding
sequence. For each category, coding and non-coding, a single EST is
chosen per cluster, in order to avoid redundancy. The matching
location also allows to determine where coding sequences start and end
in partially coding ESTs. The same annotation in FASTA headers is used
as for mRNAs. The sets of coding and non-coding ESTs are used to
perform the same computational experiments as those done with mRNA
data.
Files which already exist are reused. If an existing file is to be
recomputed, it must deleted before running the script again. For
instance, if a particular collection of mRNA or EST sequences should
be used instead of data extracted by evaluate_model, providing these
in FASTA format under the name of the mRNA file (where evaluate_model
would store the extracted data) is enough. The same procedure can be
applied to provide hand picked test data. However, in mRNA and EST
data used for test and training, evaluate_model expects annotations of
coding sequence start and stop in the header as two integer values
following the tag 'CDS:'.The first integer points to the first
nucleotide of the CDS, the second to the last. Thus the length of the
CDS is <stop> - <start> + 1. The position counting starts with 1.
=head1 DIRECTORY STRUCTURE
evaluate_model uses the same directory structure as build_model, the
root of which is given in the configuration file. From this root it
adds the subdirectory 'Evaluate', which contains all result files.
mRNA and EST data as well as test and training data files are dposited
in the data-root directory if not otherwise specified in the
configuration file.
=head1 OPTIONS AND CONFIGURATION FILE
The same command options except '-e' and variables in configuration
files as for build_model can be used. They are essentially used to
find the proper model and to create different files for different
models in a systematic manner. However, there is one additional
option:
-o <optionsfiles>
An options file contains a set of ESTScan commandline switches per
line. For each line of the file the evaluation is performed
once. The script and data files generated at each run are removed
in this mode, the results can only be collected in the Report
file. If '-o' is not specified, evaluation is performed using
ESTScan's default parameters (except for the matric to be
evaluated, of course).
Additional parameters defined in the configuration files for
evaluate_model are listed here:
* $ugdata
Name of the file(s) containing data about unigene clusters.
If this is not defined, no evaluation is currently implemented.
$filestem is used to generate many filenames. It is generated
automatically according to the tuplesize, the minmask and the
pseudocounts applied to generate them.
=head1 REQUIREMENTS
During analysis of UniGene clusters and evaluation of the generated
tables some external packages are used to collect and compare
sequences. evaluate_model relies on 'megablast' to determine where
ESTs match on full-length mRNA sequences. The 'fetch' utility is used
to find the EST and mRNA entries. This tool needs a properly indexed
version of EMBL and RefSeq flatfiles. Use 'indexer' for this. Both tools
are part of the BTLib toolset.
=head1 AUTHOR
Claudio Lottaz, SIB-ISREC, Claudio.Lottaz@isb-sib.ch
=cut
#
# End of file
#
################################################################################
|