PageRenderTime 69ms CodeModel.GetById 31ms RepoModel.GetById 1ms app.codeStats 0ms

/Bio/fm/filters/blast_xml.pm

https://github.com/sauloal/perlscripts
Perl | 944 lines | 702 code | 99 blank | 143 comment | 92 complexity | 23dc9a6449580a83065bc8687dd1b5a7 MD5 | raw file
    

  1. #!/usr/bin/perl -w
    2. 

  2. use strict;
    3. 

  3. use Bio::SearchIO; 
    4. 

  4. 

  5. $| = 1;
    6. 

  6. 

  7. my $inputFile = $ARGV[0];
    8. 

  8. my $verbose = 1;
    9. 

  9. my $Log;
    10. 

  10. 

  11. if ($verbose) { unlink("blast_xml.log"); };
    12. 

  12. 

  13. my $hypothetical = 0;
    14. 

  14. 

  15. my %priority =
    16. 

  16. (
    17. 

  17. 	"BLASTN"  => 1
    18. 

  18. );
    19. 

  19. 

  20. my %minident =
    21. 

  21. (
    22. 

  22. 	"BLASTN"  => 80
    23. 

  23. );
    24. 

  24. 

  25. my %minconsv =
    26. 

  26. (
    27. 

  27. 	"BLASTN"  => 80
    28. 

  28. );
    29. 

  29. 

  30. my %minsig =
    31. 

  31. (
    32. 

  32. 	"BLASTN"  => 1e-25
    33. 

  33. );
    34. 

  34. 

  35. my %minsize =
    36. 

  36. (
    37. 

  37. 	"BLASTN"  => 20
    38. 

  38. );
    39. 

  39. 

  40. 

  41. # $inputFile = "answer_WM276GBFF10_06_R265_c_neoformans_db_tblastx.bls";
    42. 

  42. # $inputFile = "answer_WM276GBFF10_06_R265_c_neoformans_db_blastn.bls";
    43. 

  43. # $inputFile = "answer_WM276GBFF10_06_R265_c_neoformans_db_blastn_MIX.bls";
    44. 

  44. # $inputFile = "answer_WM276GBFF10_06_R265_c_neoformans_db_blastn.bls";
    45. 

  45. $inputFile = "../blast/1_query_1.fa_1_db_db_blastn.blast";
    46. 

  46. 

  47. #&fixBlast($inputFile);
    48. 

  48. 

  49. my @answers;
    50. 

  50. my ($Gtotal, $GtotalG, %Gpositives, %Gblasts, %Gnegatives, %GnoIdent, %GnoSig, %Gshort, %GnoHit, %Gstat);
    51. 

  51. @answers    = &start($inputFile);
    52. 

  52. $Gtotal     =   $answers[0];
    53. 

  53. $GtotalG    =   $answers[1];
    54. 

  54. %Gpositives = %{$answers[2]};
    55. 

  55. %Gblasts    = %{$answers[3]};
    56. 

  56. %Gnegatives = %{$answers[4]};
    57. 

  57. %GnoIdent   = %{$answers[5]};
    58. 

  58. %GnoSig     = %{$answers[6]};
    59. 

  59. %Gshort     = %{$answers[7]};
    60. 

  60. %GnoHit     = %{$answers[8]};
    61. 

  61. %Gstat      = %{$answers[9]};
    62. 

  62. 

  63. 

  64. &printStat($Gtotal, $GtotalG, \%Gpositives, \%Gblasts, \%Gnegatives, \%GnoIdent, \%GnoSig, \%Gshort, \%GnoHit, \%Gstat);
    65. 

  65. die; 
    66. 

  66. my %GbyChrom = %{&genDB(\%Gpositives, \%Gblasts)};
    67. 

  67. &save_chromossome(\%GbyChrom, $inputFile);
    68. 

  68. &commitLog();
    69. 

  69. 

  70. sub commitLog
    71. 

  71. {
    72. 

  72. 	open FILE, ">blast_xml.log" or die "COULD NOT CREATE BLAST_XML.LOG";
    73. 

  73. 	print FILE $Log;
    74. 

  74. 	close FILE;
    75. 

  75. }
    76. 

  76. 

  77. 

  78. sub Log
    79. 

  79. {
    80. 

  80. 	if ($verbose)
    81. 

  81. 	{
    82. 

  82. 		print $_[0];
    83. 

  83. 	}
    84. 

  84. 	$Log .= $_[0];
    85. 

  85. }
    86. 

  86. 

  87. 

  88. sub fixBlast
    89. 

  89. {
    90. 

  90. 	my $input = $_[0];
    91. 

  91. 	open INFILE,  "<$input"     or die "COULD NOT OPEN INPUT FILE $input FOR FIXING\n";
    92. 

  92. 	open OUTFILE, ">$input.tmp" or die "COULD NOT OPEN INPUT FILE $input.tmp FOR FIXING\n";
    93. 

  93. 

  94. 	my $on = 0;
    95. 

  95. 	my $query = "";
    96. 

  96. 	while (<INFILE>)
    97. 

  97. 	{
    98. 

  98. 		chomp;
    99. 

  99. 		if ((/^Length=/) && ($on))
    100. 

  100. 		{
    101. 

  101. 			$query =~ s/\_+/\_/g;
    102. 

  102. 			s/\_+/\_/g;
    103. 

  103. 			print OUTFILE "Query=$query\n";
    104. 

  104. 			print OUTFILE "$_\n";
    105. 

  105. 			$on    = 0;
    106. 

  106. 			$query = "";
    107. 

  107. 		}
    108. 

  108. 		elsif (( ! $on) && ( ! (/^Query=\s*(\S+)/)))
    109. 

  109. 		{
    110. 

  110. 			s/\_+/\_/g;
    111. 

  111. 			print OUTFILE "$_\n";
    112. 

  112. 		}
    113. 

  113. 

  114. 		if ($on)
    115. 

  115. 		{
    116. 

  116. 			$query .= $_;
    117. 

  117. 		}
    118. 

  118. 		elsif (/^Query=\s*(\S+)/)
    119. 

  119. 		{
    120. 

  120. 			$query = $1;
    121. 

  121. 			$on    = 1;
    122. 

  122. 		}
    123. 

  123. 	}
    124. 

  124. 	close INFILE;
    125. 

  125. 	close OUTFILE;
    126. 

  126. 	rename $input, "$input.old";
    127. 

  127. 	rename "$input.tmp", $input;
    128. 

  128. }
    129. 

  129. 

  130. sub save_chromossome
    131. 

  131. {
    132. 

  132. 	my %db        = %{$_[0]};
    133. 

  133. 	my $inputFile =   $_[1];
    134. 

  134. 	my $log;
    135. 

  135. 

  136. 	mkdir "./xml";
    137. 

  137. 

  138. 	&Log("EXPORTING XML FILE: $inputFile.xml\n");
    139. 

  139. 	my $count = 0;
    140. 

  140. 	my %files;
    141. 

  141. 

  142. 	open  FILE, ">$inputFile.xml" or die "COULD NOT OPEN inputFile.xml XML FILE: $!";
    143. 

  143. 	print FILE  "<root id=\"$inputFile\" type=\"blast\">\n";
    144. 

  144. 	
    145. 

  145. 	foreach my $chrom (sort keys %db)
    146. 

  146. 	{
    147. 

  147. 		&Log("CHROMOSSOME $chrom\n\tSTART\n");
    148. 

  148. 		print FILE "\t<table id=\"$chrom\" type=\"chromossome\">\n";
    149. 

  149. 		my $startT = 0;
    150. 

  150. 		foreach my $start (sort { $a <=> $b } keys %{$db{$chrom}})
    151. 

  151. 		{
    152. 

  152. 			&Log("\t$start");
    153. 

  153. 			$startT++;
    154. 

  154. 			$count++;
    155. 

  155. 			my %data = %{$db{$chrom}{$start}};
    156. 

  156. 			my $out  = "\t\t<row id=\"$count\">\n";
    157. 

  157. 			$out    .= "\t\t\t<id>$count</id>\n";
    158. 

  158. 			foreach my $qual (sort keys %data)
    159. 

  159. 			{
    160. 

  160. 				if ($qual ne "chromR")
    161. 

  161. 				{
    162. 

  162. 					my $value = $data{$qual};
    163. 

  163. 					$out .= "\t\t\t<$qual>$value</$qual>\n";
    164. 

  164. 				}
    165. 

  165. 			}
    166. 

  166. 			$out .= "\t\t</row>\n";
    167. 

  167. 			print FILE $out;
    168. 

  168. 		} #end foreach my start
    169. 

  169. 		&Log("\tTOTAL $startT\n\n");
    170. 

  170. 		print FILE "\t</table>\n";
    171. 

  171. 	} #end foreach my chrom
    172. 

  172. 	print FILE "</root>";
    173. 

  173. 	close FILE;
    174. 

  174. 

  175. 	&Log("EXPORTED $count REGISTERS TO XML FILE\n");
    176. 

  176. 

  177. 	$count = 0;
    178. 

  178. 	foreach my $key (sort keys %db)
    179. 

  179. 	{
    180. 

  180. 		my $genes = (keys %{$db{$key}});
    181. 

  181. 		&Log("CHROMOSSOME $key HAS $genes GENES MAPPED\n");
    182. 

  182. 	}
    183. 

  183. 

  184. } #end sub parse snps
    185. 

  185. 

  186. 

  187. # <root id="NAME" type="[snp|gap|blast]">
    188. 

  188. #   <table id="NAME" type="[program|chromossome]">
    189. 

  189. #     <row>
    190. 

  190. #       <colum> </colum>
    191. 

  191. #     <row>
    192. 

  192. #   </table>
    193. 

  193. # </root>
    194. 

  194. 

  195. sub genDB
    196. 

  196. {
    197. 

  197. 	my @hashes  = @_;
    198. 

  198. 	my %byChrom;
    199. 

  199. 	my $countG = 0;
    200. 

  200. 	my $countU = 0;
    201. 

  201. 	my %genCoun;
    202. 

  202. 	my %faulty;
    203. 

  203. 

  204. 	print "" . @hashes . " HASHES\n";
    205. 

  205. 	foreach my $hashRef (@hashes)
    206. 

  206. 	{
    207. 

  207. 		my %hash = %{$hashRef};
    208. 

  208. 		&Log("\t THERE ARE " . (keys %hash) . " METHODS IN THIS HASH\n");
    209. 

  209. 		foreach my $method (sort keys %priority)
    210. 

  210. 		{
    211. 

  211. 			if ( ! defined %{$hash{$method}} ) { last;};
    212. 

  212. 			&Log("\t\t THE METHOD $method HAS " . (keys %{$hash{$method}}) . " GENES\n");
    213. 

  213. 			foreach my $gene (sort keys %{$hash{$method}})
    214. 

  214. 			{
    215. 

  215. 				my %data           = %{$hash{$method}{$gene}};
    216. 

  216. 				my $chromR         = $data{"chromR"};
    217. 

  217. 				my $start          = $data{"start"};
    218. 

  218. 				my $end            = $data{"end"};
    219. 

  219. 				my $sign           = $data{"sign"};
    220. 

  220. 				my $ident          = $data{"ident"};
    221. 

  221. 				my $consv          = $data{"consv"};
    222. 

  222. 				my $gaps           = $data{"gaps"};
    223. 

  223. 				my $strand         = $data{"strand"};
    224. 

  224. 				my $prop           = abs(1-($data{"Qrylen"} / $data{"Hlength"}));
    225. 

  225. 				my $size;
    226. 

  226. 				my $point;
    227. 

  227. 				if    ($strand ==  1) { $point = $start; $size = $end   - $start+1; }
    228. 

  228. 				elsif ($strand == -1) { $point = $end;   $size = $start - $end+1; }
    229. 

  229. 				else  { die "NO START POSITION FOUND"};
    230. 

  230. 

  231. 				   $data{"method"} = $method;
    232. 

  232. 				   $data{"gene"}   = $gene;
    233. 

  233. 				   $data{"size"}   = $size;
    234. 

  234. 

  235. 				if ( ! defined $byChrom{$chromR}{$point} )
    236. 

  236. 				{
    237. 

  237. 					$countU++;
    238. 

  238. 					if ( ! $genCoun{$gene} ) # to fix, order of execution shouldnt affect result
    239. 

  239. 					{
    240. 

  240. 						$byChrom{$chromR}{$point} = \%data;
    241. 

  241. 					}
    242. 

  242. 					$genCoun{$gene} += 1;
    243. 

  243. 				}
    244. 

  244. 				else
    245. 

  245. 				{
    246. 

  246. 					$faulty{$gene}++;
    247. 

  247. 

  248. 					my %dataOld   = %{$byChrom{$chromR}{$point}};
    249. 

  249. 					my $methodOld =   $dataOld{"method"};
    250. 

  250. 					my $endOld    =   $dataOld{"end"};
    251. 

  251. 					my $geneOld   =   $dataOld{"gene"};
    252. 

  252. 					my $signOld   =   $dataOld{"sign"};
    253. 

  253. 					my $identOld  =   $dataOld{"ident"};
    254. 

  254. 					my $consvOld  =   $dataOld{"consv"};
    255. 

  255. 					my $gapsOld   =   $dataOld{"gaps"};
    256. 

  256. 					my $propOld   = abs(1-($dataOld{"Qrylen"} / $dataOld{"Hlength"}));
    257. 

  257. 					my $update    = 0;
    258. 

  258. 

  259. 					if ($priority{$method} < $priority{$methodOld})
    260. 

  260. 					{
    261. 

  261. 						&Log(print "GENE $gene ON $chromR WAS UPDATED\n");
    262. 

  262. 						${$byChrom{$chromR}{$point}}[0] = \%data;
    263. 

  263. 					}
    264. 

  264. 					elsif ($priority{$methodOld} == $priority{$method})
    265. 

  265. 					{
    266. 

  266. 						my $old = 0; my $new = 0; my $draw = 0;
    267. 

  267. 						if ($signOld      < $sign)        { $old +=2; } elsif ($signOld   > $sign)           { $new +=2; } else { $draw++; };
    268. 

  268. 						if ($propOld      < $prop)        { $old++;   } elsif ($propOld   > $prop)           { $new++;   } else { $draw++; };
    269. 

  269. 						if ($gapsOld      < $gaps)        { $old++;   } elsif ($gapsOld   > $gaps)           { $new++;   } else { $draw++; };
    270. 

  270. 						if ($identOld     > $ident)       { $old++;   } elsif ($identOld  < $ident)          { $new++;   } else { $draw++; };
    271. 

  271. 						if ($consvOld     > $consv)       { $old++;   } elsif ($consvOld  < $consv)          { $new++;   } else { $draw++; };
    272. 

  272. 

  273. 						if ((20*$signOld) <      $sign)   { $old++;   } elsif ((20*$sign) >      $signOld)   { $new++;   };
    274. 

  274. 						if (( 2*$propOld) <      $prop)   { $old++;   } elsif (( 2*$prop) >      $propOld)   { $new++;   };
    275. 

  275. 						if (    $identOld < (1.1*$ident)) { $old++;   } elsif (   $ident  > (1.1*$identOld)) { $new++;   };
    276. 

  276. 						if (    $consvOld < (1.2*$consv)) { $old++;   } elsif (   $consv  > (1.2*$consvOld)) { $new++;   };
    277. 

  277. 

  278. 						if ($old == $new)                 { if (int(rand(1))) { $old++; } else { $new++; };  };
    279. 

  279. 

  280. 						if ($geneOld eq $gene)
    281. 

  281. 						{
    282. 

  282. 							&Log("DUPLICATION... FIND OUT WHERE IS IT YOUR LAZY FAT ASS\n");
    283. 

  283. 							&Log("\t\t$chromR BY ($methodOld vs $method) STARTING AT $point, ENDING AT $endOld AND $end ($old vs $new vs $draw)\n");
    284. 

  284. 							&Log("\t\tSIG\t$signOld\tvs\t$sign\n");
    285. 

  285. 							&Log("\t\tIDENT\t$identOld\tvs\t$ident\n");
    286. 

  286. 							&Log("\t\tPROP\t$propOld\tvs\t$prop\n");
    287. 

  287. 							&Log("\t\tGAPS\t$gapsOld\tvs\t$gaps\n");
    288. 

  288. 							&Log("\t\tCONSV\t$consvOld\tvs\t$consv\n");
    289. 

  289. 							&Log("\t\t$geneOld\n\t\t$gene\n\n\n");
    290. 

  290. 						}
    291. 

  291. 						elsif ($new > $old)
    292. 

  292. 						{
    293. 

  293. # 							print "KEEPING NEW DATA $old vs $new\n";
    294. 

  294. # 							print "\t\t$chromR BY ($methodOld vs $method) STARTING AT $point, ENDING AT $endOld AND $end ($old vs $new vs $draw)\n";
    295. 

  295. # 							print "\t\tSIG\t$signOld\tvs\t$sign\n";
    296. 

  296. # 							print "\t\tIDENT\t$identOld\tvs\t$ident\n";
    297. 

  297. # 							print "\t\tPROP\t$propOld\tvs\t$prop\n";
    298. 

  298. # 							print "\t\tGAPS\t$gapsOld\tvs\t$gaps\n";
    299. 

  299. # 							print "\t\tCONSV\t$consvOld\tvs\t$consv\n";
    300. 

  300. # 							print "\t\t$geneOld\n\t\t$gene\n\n\n";
    301. 

  301. 							$byChrom{$chromR}{$point} = \%data;
    302. 

  302. 						}
    303. 

  303. 						elsif  ($old > $new)
    304. 

  304. 						{
    305. 

  305. # 							print "KEEPING OLD DATA $old vs $new\n";
    306. 

  306. # 							print "\t\t$chromR BY ($methodOld vs $method) STARTING AT $point, ENDING AT $endOld AND $end ($old vs $new vs $draw)\n";
    307. 

  307. # 							print "\t\tSIG\t$signOld\tvs\t$sign\n";
    308. 

  308. # 							print "\t\tIDENT\t$identOld\tvs\t$ident\n";
    309. 

  309. # 							print "\t\tPROP\t$propOld\tvs\t$prop\n";
    310. 

  310. # 							print "\t\tGAPS\t$gapsOld\tvs\t$gaps\n";
    311. 

  311. # 							print "\t\tCONSV\t$consvOld\tvs\t$consv\n";
    312. 

  312. # 							print "\t\t$geneOld\n\t\t$gene\n\n\n";
    313. 

  313. 						}
    314. 

  314. 						else
    315. 

  315. 						{
    316. 

  316. 							&Log("SOMETHING CRAZY HAS HAPPENED: TWO GENES MAPPED TO THE SAME PLACE BY THE SAME METHOD HAVING EVERYTHING EQUAL\n");
    317. 

  317. 							&Log("\t\t$chromR BY ($methodOld vs $method) STARTING AT $point, ENDING AT $endOld AND $end ($old vs $new vs $draw)\n");
    318. 

  318. 							&Log("\t\tSIG\t$signOld\tvs\t$sign\n");
    319. 

  319. 							&Log("\t\tIDENT\t$identOld\tvs\t$ident\n");
    320. 

  320. 							&Log("\t\tPROP\t$propOld\tvs\t$prop\n");
    321. 

  321. 							&Log("\t\tGAPS\t$gapsOld\tvs\t$gaps\n");
    322. 

  322. 							&Log("\t\tCONSV\t$consvOld\tvs\t$consv\n");
    323. 

  323. 							&Log("\t\t$geneOld\n\t\t$gene\n\n\n");
    324. 

  324. 							$byChrom{$chromR}{$point} = \%data;
    325. 

  325. 						} # end if ($new > $old)
    326. 

  326. 					} # end if ($priority{$method} < $priority{$methodOld})
    327. 

  327. 				} # end if ( defined @{$byChrom{$chromR}{$point}} )
    328. 

  328. 				$countG++;
    329. 

  329. 			} # end foreach my $gene (sort keys %{$hash{$method}})
    330. 

  330. 		} # end foreach my $method (sort keys %hash)
    331. 

  331. 	} # end foreach my $hashRef (@hashes)
    332. 

  332. 

  333. 	&Log("$countG GENES OF INPUT BEING $countU UNIQUES\n");
    334. 

  334. 	return (\%byChrom);
    335. 

  335. }
    336. 

  336. 

  337. 

  338. sub getChar
    339. 

  339. {
    340. 

  340. 	my $method     = $_[0];
    341. 

  341. 	my $query_name = $_[1];
    342. 

  342. 	my $valid      = $_[2];
    343. 

  343. 	my %data = ();
    344. 

  344. 

  345. 	if (exists $Gpositives{$method}{$query_name})
    346. 

  346. 	{
    347. 

  347. 		%data = %{$Gpositives{$method}{$query_name}};
    348. 

  348. 	}
    349. 

  349. 	elsif (exists $Gblasts{$method}{$query_name})
    350. 

  350. 	{
    351. 

  351. 		%data = %{$Gblasts{$method}{$query_name}};
    352. 

  352. 	}
    353. 

  353. 	elsif (exists $Gnegatives{$method}{$query_name}) 
    354. 

  354. 	{
    355. 

  355. 		%data = %{$Gnegatives{$method}{$query_name}};
    356. 

  356. 	}
    357. 

  357. 	else
    358. 

  358. 	{
    359. 

  359. # 		%data = %{$Gnegatives{$method}{$query_name}};
    360. 

  360. # 		print "*"x20 . "\nTHERE'S NO DATA FOR $query_name IN METHOD $method\n" . "*"x20 . "\n";
    361. 

  361. 	}
    362. 

  362. 

  363. 	return \%data;
    364. 

  364. }
    365. 

  365. 

  366. 

  367. sub start
    368. 

  368. {
    369. 

  369. 	my $inputFile = $_[0];
    370. 

  370. 	if ( -f $inputFile ) { &Log("LOADING FILE $inputFile\n"); } else { die "COULD NOT OPEN $inputFile: $!"};
    371. 

  371. 	my $in = new Bio::SearchIO(-format => 'blast', 
    372. 

  372. 				   -file   => $inputFile);
    373. 

  373. 	$| =0;
    374. 

  374. 	my %positives = ();
    375. 

  375. 	my %blasts    = ();
    376. 

  376. 	my %negatives = ();
    377. 

  377. 	my %noIdent   = ();
    378. 

  378. 	my %noSig     = ();
    379. 

  379. 	my %short     = ();
    380. 

  380. 	my %noHit     = ();
    381. 

  381. 	my %stat      = ();
    382. 

  382. 

  383. 	my $Positives = 0;
    384. 

  384. 	my $Negatives = 0;
    385. 

  385. 	my $Blasts    = 0;
    386. 

  386. 	my $NoHits    = 0;
    387. 

  387. 

  388. 	my $total     = 0;
    389. 

  389. 	my $totalG    = 0;
    390. 

  390. # 	my %totalG;
    391. 

  391. # 	my %POSG;
    392. 

  392. # 	my %NEGG;
    393. 

  393. # 	my %NOHITG;
    394. 

  394. 

  395. 	while( my $result = $in->next_result )
    396. 

  396. 	{
    397. 

  397. 	# 	print "METHOD " . $result->algorithm . "\n";
    398. 

  398. 		my $method   = $result->algorithm;
    399. 

  399. 		my $minSize  = $minsize{$method};
    400. 

  400. 		my $minSig   = $minsig{$method};
    401. 

  401. 		my $minIdent = $minident{$method};
    402. 

  402. 		my $minConsv = $minconsv{$method};
    403. 

  403. 	
    404. 

  404. # 		print ".";
    405. 

  405. 		#   print	"Query Name   = " . $result->query_name   . "\n";
    406. 

  406. 		#   print	"Query Length = " . $result->query_length . "\n";
    407. 

  407. 		my $num_hits = $result->num_hits;
    408. 

  408. 		#   print	"Query Hits   = " . $result->num_hits     . "\n";
    409. 

  409. 		my $hitC = 0;
    410. 

  410. 		
    411. 

  411. 		$totalG++;
    412. 

  412. # 		$totalG{$result->query_name}++;
    413. 

  413. 

  414. 		if ( $result->num_hits )
    415. 

  415. 		{
    416. 

  416. 			while( my $hit = $result->next_hit )
    417. 

  417. 			{
    418. 

  418. 				$hitC++;
    419. 

  419. 				my $hspC = 0;
    420. 

  420. 				#   print "\tHit #$hitC/$num_hits   = " . $hit->name             . "\n",
    421. 

  421. 				# 	"\tHit Significance = " . $hit->significance     . "\n",
    422. 

  422. 				# 	"\tHit hsps         = " . $hit->num_hsps         . "\n";
    423. 

  423. 				while( my $hsp = $hit->next_hsp )
    424. 

  424. 				{
    425. 

  425. 					$hspC++;
    426. 

  426. 					if( $hsp->length('total') > $minSize )
    427. 

  427. 					{
    428. 

  428. 						if (( $hsp->percent_identity >= ($minIdent/100) ) || ($hsp->frac_conserved >= ($minConsv/100) ))
    429. 

  429. 						{
    430. 

  430. 							if ( $hit->significance <= $minSig )
    431. 

  431. 							{
    432. 

  432. #								if (($hsp->rank == 1) && ($hspC == 1) && ($hitC == 1))
    433. 

  433. #								{
    434. 

  434. 									$stat{$method}{$hsp->evalue}++; $total++;
    435. 

  435. #									if (($result->num_hits == 1) && ($hit->num_hsps == 1))
    436. 

  436. #									{
    437. 

  437. # 										print ":";
    438. 

  438. 										$positives{$method}{$result->query_name}{"ident"}   = &roundFrac($hsp->frac_identical *100);
    439. 

  439. 	 									$positives{$method}{$result->query_name}{"consv"}   = &roundFrac($hsp->frac_conserved *100);
    440. 

  440. 										$positives{$method}{$result->query_name}{"sign"}    = $hit->significance;
    441. 

  441. 										$positives{$method}{$result->query_name}{"Hlength"} = $hsp->length('hit');
    442. 

  442. 										$positives{$method}{$result->query_name}{"Qlength"} = $hsp->length('query');
    443. 

  443. 										$positives{$method}{$result->query_name}{"Tlength"} = $hsp->length('total');
    444. 

  444. 										$positives{$method}{$result->query_name}{"Qrylen"}  = $result->query_length;
    445. 

  445. 										$positives{$method}{$result->query_name}{"start"}   = $hsp->start('hit');
    446. 

  446. 										$positives{$method}{$result->query_name}{"end"}     = $hsp->end('hit');
    447. 

  447. 										$positives{$method}{$result->query_name}{"strand"}  = $hsp->strand('hit');
    448. 

  448. 										my $name = $hit->name; $name =~ s/lcl\|//;
    449. 

  449. 	 									$positives{$method}{$result->query_name}{"chromR"}  = $name;
    450. 

  450. 										$positives{$method}{$result->query_name}{"gaps"}    = $hsp->gaps;
    451. 

  451. # 										$Positives++;
    452. 

  452. # 										$POSG{$result->query_name}++;
    453. 

  453. #									}
    454. 

  454. #									else
    455. 

  455. #									{
    456. 

  456. # 										print ".";
    457. 

  457. #										$blasts{$method}{$result->query_name}{"ident"}   = &roundFrac($hsp->frac_identical *100);
    458. 

  458. #										$blasts{$method}{$result->query_name}{"consv"}   = &roundFrac($hsp->frac_conserved *100);
    459. 

  459. #										$blasts{$method}{$result->query_name}{"sign"}    = $hit->significance;
    460. 

  460. #										$blasts{$method}{$result->query_name}{"Hlength"} = $hsp->length('hit');
    461. 

  461. #										$blasts{$method}{$result->query_name}{"Qlength"} = $hsp->length('query');
    462. 

  462. #										$blasts{$method}{$result->query_name}{"Tlength"} = $hsp->length('total');
    463. 

  463. #										$blasts{$method}{$result->query_name}{"Qrylen"}  = $result->query_length;
    464. 

  464. #										$blasts{$method}{$result->query_name}{"start"}   = $hsp->start('hit');
    465. 

  465. #										$blasts{$method}{$result->query_name}{"end"}     = $hsp->end('hit');
    466. 

  466. #										$blasts{$method}{$result->query_name}{"strand"}  = $hsp->strand('hit');
    467. 

  467. #										my $name = $hit->name; $name =~ s/lcl\|//;
    468. 

  468. #										$blasts{$method}{$result->query_name}{"chromR"}  = $name;
    469. 

  469. #										$blasts{$method}{$result->query_name}{"gaps"}    = $hsp->gaps;
    470. 

  470. # 										$Blasts++;
    471. 

  471. # 										$POSG{$result->query_name}++;
    472. 

  472. #									} #end if else num hits = 1
    473. 

  473. #								}; # end if hank = 1
    474. 

  474. 

  475. 							#       print	"\t\tHsp Length       = " . $hsp->length('total')  . "\n",
    476. 

  476. 							# 		"\t\tHsp Percent_id   = " . $hsp->percent_identity . "\n",
    477. 

  477. 							# 		"\t\tHsp e-value      = " . $hsp->evalue           . "\n",
    478. 

  478. 							# 		"\t\tHsp Frac_Identic = " . $hsp->frac_identical   . "\n",
    479. 

  479. 							# 		"\t\tHsp Frac_conserv = " . $hsp->frac_conserved   . "\n",
    480. 

  480. 							# 		"\t\tHsp Rank         = " . $hsp->rank             . "\n",
    481. 

  481. 							# 		"\t\tHsp Gaps         = " . $hsp->gaps             . "\n",
    482. 

  482. 							# 		"\t\tHsp Range      @ = " . join ("\t", $hsp->range('hit'))     . "\n",
    483. 

  483. 							# 
    484. 

  484. 							# 		"\t\tHsp Query Strand = " . $hsp->strand('query')  . "\n",
    485. 

  485. 							# 		"\t\tHsp Query Start  = " . $hsp->start('query')   . "\n",
    486. 

  486. 							# 		"\t\tHsp Query End    = " . $hsp->end('query')     . "\n",
    487. 

  487. 							# 		"\t\tHsp Query Length = " . $hsp->length('query')  . "\n",
    488. 

  488. 							# # 		"\t\tHsp Query String = " . $hsp->query_string 	   . "\n",
    489. 

  489. 							# 
    490. 

  490. 							# 		"\t\tHsp Hit Strand   = " . $hsp->strand('hit')    . "\n",
    491. 

  491. 							# 		"\t\tHsp Hit Start    = " . $hsp->start('hit')     . "\n",
    492. 

  492. 							# 		"\t\tHsp Hit End      = " . $hsp->end('hit')       . "\n",
    493. 

  493. 							# 		"\t\tHsp Hit Length   = " . $hsp->length('hit')    . "\n",
    494. 

  494. 							# # 		"\t\tHsp Hit String   = " . $hsp->hit_string       . "\n",
    495. 

  495. 							# # 		"\t\tHsp Homology Str = " . $hsp->homology_string  . "\n",
    496. 

  496. 									;
    497. 

  497. 							} # end if sig > 1e-20
    498. 

  498. 							else
    499. 

  499. 							{
    500. 

  500. 								if (($hsp->rank == 1) && ($hspC == 1) && ($hitC == 1))
    501. 

  501. 								{
    502. 

  502. # 									print ",";
    503. 

  503. 									$noSig{$method}{$result->query_name}++;
    504. 

  504. 									$negatives{$method}{$result->query_name}{"ident"}   = &roundFrac($hsp->frac_identical *100);
    505. 

  505. 									$negatives{$method}{$result->query_name}{"consv"}   = &roundFrac($hsp->frac_conserved *100);
    506. 

  506. 									$negatives{$method}{$result->query_name}{"sign"}    = $hit->significance;
    507. 

  507. 									$negatives{$method}{$result->query_name}{"Hlength"} = $hsp->length('hit');
    508. 

  508. 									$negatives{$method}{$result->query_name}{"Qlength"} = $hsp->length('query');
    509. 

  509. 									$negatives{$method}{$result->query_name}{"Tlength"} = $hsp->length('total');
    510. 

  510. 									$negatives{$method}{$result->query_name}{"Qrylen"}  = $result->query_length;
    511. 

  511. 									$negatives{$method}{$result->query_name}{"start"}   = $hsp->start('hit');
    512. 

  512. 									$negatives{$method}{$result->query_name}{"end"}     = $hsp->end('hit');
    513. 

  513. 									$negatives{$method}{$result->query_name}{"strand"}  = $hsp->strand('hit');
    514. 

  514. 									$negatives{$method}{$result->query_name}{"hits"}    = $result->num_hits;
    515. 

  515. 									my $name = $hit->name; $name =~ s/lcl\|//;
    516. 

  516. 									$negatives{$method}{$result->query_name}{"chromR"}  = $name;
    517. 

  517. 									$negatives{$method}{$result->query_name}{"gaps"}    = $hsp->gaps;
    518. 

  518. # 									$Negatives++;
    519. 

  519. # 									$NEGG{$result->query_name}++;
    520. 

  520. 								};
    521. 

  521. 							};
    522. 

  522. 						} # end if ident < 80
    523. 

  523. 						else
    524. 

  524. 						{
    525. 

  525. 							if (($hsp->rank == 1) && ($hspC == 1) && ($hitC == 1))
    526. 

  526. 							{
    527. 

  527. # 								print ",";
    528. 

  528. 								$noIdent{$method}{$result->query_name}++; 
    529. 

  529. 								$negatives{$method}{$result->query_name}{"ident"}   = &roundFrac($hsp->frac_identical *100);
    530. 

  530. 								$negatives{$method}{$result->query_name}{"consv"}   = &roundFrac($hsp->frac_conserved *100);
    531. 

  531. 								$negatives{$method}{$result->query_name}{"sign"}    = $hit->significance;
    532. 

  532. 								$negatives{$method}{$result->query_name}{"Hlength"} = $hsp->length('hit');
    533. 

  533. 								$negatives{$method}{$result->query_name}{"Qlength"} = $hsp->length('query');
    534. 

  534. 								$negatives{$method}{$result->query_name}{"Tlength"} = $hsp->length('total');
    535. 

  535. 								$negatives{$method}{$result->query_name}{"Qrylen"}  = $result->query_length;
    536. 

  536. 								$negatives{$method}{$result->query_name}{"start"}   = $hsp->start('hit');
    537. 

  537. 								$negatives{$method}{$result->query_name}{"end"}     = $hsp->end('hit');
    538. 

  538. 								$negatives{$method}{$result->query_name}{"strand"}  = $hsp->strand('hit');
    539. 

  539. 								$negatives{$method}{$result->query_name}{"hits"}    = $result->num_hits;
    540. 

  540. 								my $name = $hit->name; $name =~ s/lcl\|//;
    541. 

  541. 								$negatives{$method}{$result->query_name}{"chromR"}  = $name;
    542. 

  542. 								$negatives{$method}{$result->query_name}{"gaps"}    = $hsp->gaps;
    543. 

  543. # 								$Negatives++;
    544. 

  544. # 								$NEGG{$result->query_name}++;
    545. 

  545. 							};
    546. 

  546. 						};
    547. 

  547. 					} # end if length < 50
    548. 

  548. 					else 
    549. 

  549. 					{
    550. 

  550. 						if (($hsp->rank == 1) && ($hspC == 1) && ($hitC == 1))
    551. 

  551. 						{
    552. 

  552. # 							print ",";
    553. 

  553. 							$short{$method}{$result->query_name}++;
    554. 

  554. 							$negatives{$method}{$result->query_name}{"ident"}   = &roundFrac($hsp->frac_identical *100);
    555. 

  555. 							$negatives{$method}{$result->query_name}{"consv"}   = &roundFrac($hsp->frac_conserved *100);
    556. 

  556. 							$negatives{$method}{$result->query_name}{"sign"}    = $hit->significance;
    557. 

  557. 							$negatives{$method}{$result->query_name}{"Hlength"} = $hsp->length('hit');
    558. 

  558. 							$negatives{$method}{$result->query_name}{"Qlength"} = $hsp->length('query');
    559. 

  559. 							$negatives{$method}{$result->query_name}{"Tlength"} = $hsp->length('total');
    560. 

  560. 							$negatives{$method}{$result->query_name}{"Qrylen"}  = $result->query_length;
    561. 

  561. 							$negatives{$method}{$result->query_name}{"start"}   = $hsp->start('hit');
    562. 

  562. 							$negatives{$method}{$result->query_name}{"end"}     = $hsp->end('hit');
    563. 

  563. 							$negatives{$method}{$result->query_name}{"strand"}  = $hsp->strand('hit');
    564. 

  564. 							$negatives{$method}{$result->query_name}{"hits"}    = $result->num_hits;
    565. 

  565. 							my $name = $hit->name; $name =~ s/lcl\|//;
    566. 

  566. 							$negatives{$method}{$result->query_name}{"chromR"}  = $name;
    567. 

  567. 							$negatives{$method}{$result->query_name}{"gaps"}    = $hsp->gaps;
    568. 

  568. # 							$Negatives++;
    569. 

  569. # 							$NEGG{$result->query_name}++;
    570. 

  570. 						};
    571. 

  571. 					};
    572. 

  572. 				}; # end while hsp
    573. 

  573. 			}; # end while result
    574. 

  574. 		} # end if hits
    575. 

  575. 		else
    576. 

  576. 		{
    577. 

  577. # 			print " ";
    578. 

  578. 			$noHit{$method}{$result->query_name}++;
    579. 

  579. # 			$NoHits++;
    580. 

  580. # 			$NOHITG{$result->query_name}++;
    581. 

  581. 		};
    582. 

  582. 	}; # end if $result->num_hits
    583. 

  583. 

  584. 	$totalG = ($totalG/(keys %stat));
    585. 

  585. # 	foreach my $name (keys %POSG)
    586. 

  586. # 	{
    587. 

  587. # 		if ($POSG{$name} > 1)
    588. 

  588. # 		{
    589. 

  589. # 			print "*"x20 . "\n$name IS DUPLICATED ON POSITIVE\n" . "*"x20;
    590. 

  590. # 		}
    591. 

  591. # 	}
    592. 

  592. # 
    593. 

  593. # 	foreach my $name (keys %NEGG)
    594. 

  594. # 	{
    595. 

  595. # 		if ($NEGG{$name} > 1)
    596. 

  596. # 		{
    597. 

  597. # 			print "*"x20 . "\n$name IS DUPLICATED ON NEGATIVE\n" . "*"x20;
    598. 

  598. # 		}
    599. 

  599. # 	}
    600. 

  600. # 
    601. 

  601. # 	foreach my $name (keys %NOHITG)
    602. 

  602. # 	{
    603. 

  603. # 		if ($NOHITG{$name} > 1)
    604. 

  604. # 		{
    605. 

  605. # 			print "*"x20 . "\n$name IS DUPLICATED ON NO HIT\n" . "*"x20;
    606. 

  606. # 		}
    607. 

  607. # 	}
    608. 

  608. 

  609. # 	my $dv  = $Positives+$Blasts+$Negatives+$NoHits;
    610. 

  610. # 	my $dv2 = (keys %totalG);
    611. 

  611. # 	my $dv3 = (keys %POSG);
    612. 

  612. # 	my $dv4 = (keys %NEGG);
    613. 

  613. # 	my $dv5 = (keys %NOHITG);
    614. 

  614. # 	print "POS $Positives + BLAST $Blasts = " . ($Positives+$Blasts) . " ($dv3)\n";
    615. 

  615. # 	print "NEG $Negatives + NOHIT $NoHits = " . ($Negatives+$NoHits) . " ($dv4 | $dv5)\n";
    616. 

  616. # 	print "VALID " . ($Positives+$Blasts) . " + INVALID " . ($Negatives+$NoHits) . " = $dv\n";
    617. 

  617. # 	print "TOTAL $totalG (DV1 $dv | DV2 $dv2)\n";
    618. 

  618. 	$| =1;
    619. 

  619. 	&Log("$inputFile LOADED\n");
    620. 

  620. 	return ($total, $totalG, \%positives, \%blasts, \%negatives, \%noIdent, \%noSig, \%short, \%noHit, \%stat)
    621. 

  621. }
    622. 

  622. 

  623. sub printStat
    624. 

  624. {
    625. 

  625. 	my $total     =   $_[0];
    626. 

  626. 	my $totalG    =   $_[1];
    627. 

  627. 	my %positives = %{$_[2]};
    628. 

  628. 	my %blasts    = %{$_[3]};
    629. 

  629. 	my %negatives = %{$_[4]};
    630. 

  630. 	my %noIdent   = %{$_[5]};
    631. 

  631. 	my %noSig     = %{$_[6]};
    632. 

  632. 	my %short     = %{$_[7]};
    633. 

  633. 	my %noHit     = %{$_[8]};
    634. 

  634. 	my %stat      = %{$_[9]};
    635. 

  635. 

  636. 	foreach my $method (sort {$a cmp $b} keys %stat)
    637. 

  637. 	{
    638. 

  638. 		my $minSize  = $minsize{$method};
    639. 

  639. 		my $minSig   = $minsig{$method};
    640. 

  640. 		my $minIdent = $minident{$method};
    641. 

  641. 		my $minConsv = $minconsv{$method};
    642. 

  642. 		my @resume   = (0)x4;
    643. 

  643. 		my $total2;
    644. 

  644. 		
    645. 

  645. 		foreach my $key (sort {$a <=> $b} keys %{$stat{$method}})
    646. 

  646. 		{
    647. 

  647. 			if ($key == 0.0)
    648. 

  648. 			{
    649. 

  649. 				$resume[0] += $stat{$method}{$key};
    650. 

  650. 				$total2    += $stat{$method}{$key};
    651. 

  651. 			}
    652. 

  652. 			elsif (($key > 0) && ($key <= 1e-35))
    653. 

  653. 			{
    654. 

  654. 				$resume[1] += $stat{$method}{$key};
    655. 

  655. 				$total2    += $stat{$method}{$key};
    656. 

  656. 			}
    657. 

  657. 			elsif (($key > 1e-35) && ($key <= 1e-30))
    658. 

  658. 			{
    659. 

  659. 				$resume[2] += $stat{$method}{$key};
    660. 

  660. 				$total2    += $stat{$method}{$key};
    661. 

  661. 			}
    662. 

  662. 			elsif (($key > 1e-30) && ($key <= $minSig))
    663. 

  663. 			{
    664. 

  664. 				$resume[3] += $stat{$method}{$key};
    665. 

  665. 				$total2    += $stat{$method}{$key};
    666. 

  666. 			}
    667. 

  667. 			elsif ($key > $minSig)
    668. 

  668. 			{
    669. 

  669. 				$resume[4] += $stat{$method}{$key};
    670. 

  670. 				$total2    += $stat{$method}{$key};
    671. 

  671. 			}
    672. 

  672. 			else
    673. 

  673. 			{
    674. 

  674. 				die "ERROR COUNTING KEY: $key";
    675. 

  675. 			}
    676. 

  676. 		} # end foreach my key
    677. 

  677. 		my $cum  = 0;
    678. 

  678. 		my $cumP = 0;
    679. 

  679. 		if ($total && $total2)
    680. 

  680. 		{
    681. 

  681. 			my $noSig     = (keys %{$noSig{$method}});
    682. 

  682. 			my $noIdent   = (keys %{$noIdent{$method}});
    683. 

  683. 			my $short     = (keys %{$short{$method}});
    684. 

  684. 			my $noHit     = (keys %{$noHit{$method}});
    685. 

  685. 			my $positives = (keys %{$positives{$method}});
    686. 

  686. 			my $blasts    = (keys %{$blasts{$method}});
    687. 

  687. 			my $negatives = (keys %{$negatives{$method}});
    688. 

  688. 

  689. 			my $invalid   = $noSig+$noIdent+$short+$noHit;
    690. 

  690. #			if ($invalid != ($totalG-$total2))   { die "INVALIDS ($invalid) IS DIFFERENT FROM THE SUBTRATION OF TOTAL ($totalG) MINUS IDENTIFIED ($total2)." };
    691. 

  691. #			if ($invalid != ($negatives+$noHit)) { die "INVALIDS ($invalid) IS DIFFERENT FROM NEGATIVES ($negatives) PLUS NOHITS ($noHit) [" . ($negatives + $noHit) . "]." };
    692. 

  692. 	
    693. 

  693. 			&Log("FILENAME         : $inputFile\n");
    694. 

  694. 			&Log("METHOD           : $method\n");
    695. 

  695. 			&Log("TOTAL GLOBAL     : $totalG\n");
    696. 

  696. 			&Log("TOTAL INDENTIFIED: $total2"    . "\t( "       . &percentage($total2   ,$totalG) . "% )\n");
    697. 

  697. 			&Log("   UNIQUES       : $positives" . "\t( "       . &percentage($positives,$totalG) . "% )\n");
    698. 

  698. 			&Log("   MULTI RESULTS : $blasts"    . "\t( "       . &percentage($blasts   ,$totalG) . "% )\n");
    699. 

  699. 			&Log("INVALIDS         : $invalid"   . "\t( "       . &percentage($invalid  ,$totalG) . "% )\n");
    700. 

  700. 			&Log("   NEGATIVES     : $negatives" . "\t( "       . &percentage($negatives,$totalG) . "% )\n");
    701. 

  701. 			&Log("   NO HITS       : $noHit"     . "\t( "       . &percentage($noHit    ,$totalG) . "% )\n");
    702. 

  702. 

  703. 			&Log("\nRESUME \n");
    704. 

  704. 			$cum += $resume[0];
    705. 

  705. 

  706. 			&Log("   VALIDS  $total2  ("                        . &percentage($total2,$totalG)    . "%)\n"); $cum += $noSig;
    707. 

  707. 			&Log("      E-VALUE == 0.0             : "          . $resume[0]                      . "\t( "  . &percentage($resume[0],$totalG) . "% CUM $cum [ "  . &percentage($cum,$totalG). "%] )\n"); $cum += $resume[1];
    708. 

  708. 			&Log("      0.0     < E-VALUE <= 1e-35 : "          . $resume[1]                      . "\t( "  . &percentage($resume[1],$totalG) . "% CUM $cum [ "  . &percentage($cum,$totalG). "%] )\n"); $cum += $resume[2];
    709. 

  709. 			&Log("      1e-35   < E-VALUE <= 1e-30 : "          . $resume[2]                      . "\t( "  . &percentage($resume[2],$totalG) . "% CUM $cum [ "  . &percentage($cum,$totalG). "%] )\n"); $cum += $resume[3];
    710. 

  710. 			&Log("      1e-30   < E-VALUE <= $minSig : "        . $resume[3]                      . "\t( "  . &percentage($resume[3],$totalG) . "% CUM $cum [ "  . &percentage($cum,$totalG). "%] )\n"); $cum += $resume[4];
    711. 

  711. 			&Log("                E-VALUE >  $minSig : "        . $resume[4]                      . "\t( "  . &percentage($resume[3],$totalG) . "% CUM $cum [ "  . &percentage($cum,$totalG). "%] )\n");
    712. 

  712. 			&Log("\n");
    713. 

  713. 			&Log("   INVALIDS  $invalid  ("                     . &percentage($invalid,$totalG)   . "%)\n"); $cum += $noSig;
    714. 

  714. 			&Log("      E-VALUE  > " . $minSig       . "\t\t: " . $noSig                          . "\t( "  . &percentage($noSig  ,$totalG)   . "% CUM $cum [ " . &percentage($cum,$totalG) . "%] )\n"); $cum += $noIdent;
    715. 

  715. 			&Log("      ID < $minIdent% & SIM < $minConsv%\t: " . $noIdent                        . "\t( "  . &percentage($noIdent,$totalG)   . "% CUM $cum [ " . &percentage($cum,$totalG) . "%] )\n"); $cum += $short;
    716. 

  716. 			&Log("      LENGTH   < " . $minSize      . "\t\t: " . $short                          . "\t( "  . &percentage($short  ,$totalG)   . "% CUM $cum [ " . &percentage($cum,$totalG) . "%] )\n"); $cum += $noHit;
    717. 

  717. 			&Log("      NO HIT   \t" .   ""          . "\t\t: " . $noHit                          . "\t( "  . &percentage($noHit  ,$totalG)   . "% CUM $cum [ " . &percentage($cum,$totalG) . "%] )\n");
    718. 

  718. 		
    719. 

  719. 			&Log("\n\n" . "-"x50 . "\n\n");
    720. 

  720. 			&Log("E-VALUE  > "         . "$minSig  \t("         . (keys %{$noSig{$method}})       . ")    \t:\n"); &printHash(\%noSig,  $method);
    721. 

  721. 			&Log("ID < $minIdent% & SIM < $minConsv\t("         . (keys %{$noIdent{$method}})     . ")" ."\t:\n"); &printHash(\%noIdent,$method);
    722. 

  722. 			&Log("LENGTH   < "         . "$minSize \t("         . (keys %{$short{$method}})       . ")    \t:\n"); &printHash(\%short,  $method);
    723. 

  723. 			&Log("NO HIT   \t(" .  ""                           . (keys %{$noHit{$method}})       . ")    \t:\n"); &printHash(\%noHit,  $method);
    724. 

  724. 			&Log("\n\n" . "="x50 . "\n\n");
    725. 

  725. 		}
    726. 

  726. 		else
    727. 

  727. 		{
    728. 

  728. 			&Log("COULD NOT RETRIEVE RESULTS\n\n");
    729. 

  729. 		}
    730. 

  730. 	} # end foreach my method
    731. 

  731. 

  732. 	&printHashSumary(\%noSig,\%noIdent,\%short,\%noHit);
    733. 

  733. }
    734. 

  734. 

  735. 

  736. sub getSumaryStr
    737. 

  737. {
    738. 

  738. 	my @hashes  = @_;
    739. 

  739. 	my %sumary;
    740. 

  740. 	my %methodC;
    741. 

  741. 	foreach my $hashRef (@hashes)
    742. 

  742. 	{
    743. 

  743. 		my %hash = %{$hashRef};
    744. 

  744. 		foreach my $method (sort keys %hash)
    745. 

  745. 		{
    746. 

  746. 			foreach my $gene (sort keys %{$hash{$method}})
    747. 

  747. 			{
    748. 

  748. 				my $data  = &getCharStr($method,$gene);
    749. 

  749. 				if (! $data ) {die "NO DATA RETRIEVED"; };
    750. 

  750. 				my $value      = $hash{$method}{$gene};
    751. 

  751. 				$sumary{$gene}{"count"} += 1;
    752. 

  752. 				push(@{$sumary{$gene}{"char"}}, $data);
    753. 

  753. 				$methodC{$method}++;
    754. 

  754. 			}
    755. 

  755. 		}
    756. 

  756. 	}
    757. 

  757. 	return (\%sumary, \%methodC);
    758. 

  758. }
    759. 

  759. 

  760. sub printHashSumary
    761. 

  761. {
    762. 

  762. 	my @result  = &getSumaryStr(@_);
    763. 

  763. 	my %sumary  = %{$result[0]};
    764. 

  764. 	my %methodC = %{$result[1]};
    765. 

  765. 	my %stat;
    766. 

  766. 

  767. 	if ((keys %methodC) > 1)
    768. 

  768. 	{
    769. 

  769. 		&Log("\n\n" . "#"x50 . "\n" . "#"x50 . "\n" . "#"x50 . "\n");
    770. 

  770. 		&Log("NOT IDENTIFIED GENES : \n");
    771. 

  771. 		&Log("\n\n" . "#"x50 . "\n" . "#"x50 . "\n" . "#"x50 . "\n");
    772. 

  772. 		my $countRec  = 0;
    773. 

  773. 		my $countLost = 0;
    774. 

  774. 	
    775. 

  775. 		my $outRec  = "";
    776. 

  776. 		my $outLost = "";
    777. 

  777. 	
    778. 

  778. 		foreach my $gene (sort keys %sumary)
    779. 

  779. 		{
    780. 

  780. 			my $value =   $sumary{$gene}{"count"};
    781. 

  781. 			my @data  = @{$sumary{$gene}{"char"}};
    782. 

  782. 			my $valid = 1;
    783. 

  783. 

  784. 			if ( ( ! $hypothetical ) && ( $gene =~ /hypothetical.protein/sig ) && ( ! ($gene =~ /conserved/sig ) ) ) { $valid = 0; };
    785. 

  785. 

  786. 			if    ($value == 1) # found in only one of the lists, so, was a match in the other
    787. 

  787. 			{
    788. 

  788. 				my $data  = $data[0];
    789. 

  789. 				$countRec++;
    790. 

  790. 				if ($valid) { $outRec  .= "\t$countRec\t$data\t$gene\n"};
    791. 

  791. 				$stat{"rec"}++;
    792. 

  792. 			}
    793. 

  793. 			elsif ($value > 1) # found in both lists, so was a mismatch every time
    794. 

  794. 			{
    795. 

  795. 				$countLost++;
    796. 

  796. 				$outLost .= "\t$countLost" if ($valid);
    797. 

  797. 				my $second = 0;
    798. 

  798. 				foreach my $data (@data)
    799. 

  799. 				{
    800. 

  800. 					if ($valid)
    801. 

  801. 					{
    802. 

  802. 						if ($second++) { $outLost .= "\t"; };
    803. 

  803. 						$outLost .= "\t$data\t$gene\n";
    804. 

  804. 					}
    805. 

  805. 				}
    806. 

  806. 				$stat{"lost"}++;
    807. 

  807. 			}
    808. 

  808. 		}
    809. 

  809. 	
    810. 

  810. 		my $total = $countRec+$countLost;
    811. 

  811. 	
    812. 

  812. 		&Log(  
    813. 

  813. 		"XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX\n",
    814. 

  814. 		"X              LOST             X\n",
    815. 

  815. 		"X===============================X\n",
    816. 

  816. 		"X    LOST      BEFORE  AFTER    X\n");
    817. 

  817. 			foreach my $prog (keys %methodC)
    818. 

  818. 			{
    819. 

  819. 				my $value = $methodC{$prog};
    820. 

  820. 				&Log("X    $prog\t$value\t$countLost\tX\n");
    821. 

  821. 			}
    822. 

  822. 		&Log(
    823. 

  823. 		"X===============================X\n",
    824. 

  824. 		"X           RECOVERED           X\n",
    825. 

  825. 		"X===============================X\n",
    826. 

  826. 		"X    LOST              AFTER    X\n");
    827. 

  827. 			foreach my $prog (keys %methodC)
    828. 

  828. 			{
    829. 

  829. 				my $value = $methodC{$prog};
    830. 

  830. 				my $minus = $value - $countLost;
    831. 

  831. 				&Log("X    $prog\t\t$minus\tX\n");
    832. 

  832. 			}
    833. 

  833. 		&Log(
    834. 

  834. 		"X===============================X\n",
    835. 

  835. 		"X             TOTAL             X\n",
    836. 

  836. 		"X===============================X\n",
    837. 

  837. 		"X    LOST	    	$countLost\tX\n",
    838. 

  838. 		"X    RECOVERED	    	$countRec\tX\n",
    839. 

  839. 		"X    TOTAL	    	$total\tX\n",
    840. 

  840. 		"XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX\n");
    841. 

  841. 	
    842. 

  842. 	
    843. 

  843. 		&Log("\n\tTOTAL $total\n");
    844. 

  844. 		&Log("\tRECUPERED $countRec\n");
    845. 

  845. 		&Log("\tCOUNT\tMETHOD\tHITS\tIDENT\tCONSV\tSIGN\tHLENGTH\tQLENGTH\tTLENGTH\tSTART\tEND\tSTRAND\tNAME\n");
    846. 

  846. 		&Log($outRec);
    847. 

  847. 		&Log("\n" . "-"x50 . "\n");
    848. 

  848. 		&Log("\n\tLOST $countLost\n");
    849. 

  849. 		&Log("\tCOUNT\tMETHOD\tHITS\tIDENT\tCONSV\tSIGN\tHLENGTH\tQLENGTH\tTLENGTH\tSTART\tEND\tSTRAND\tNAME\n");
    850. 

  850. 		&Log($outLost);
    851. 

  851. 		&Log("\n\n");
    852. 

  852. 	} #end if keys methodC > 1
    853. 

  853. }
    854. 

  854. 

  855. 

  856. 

  857. sub printHash
    858. 

  858. {
    859. 

  859. 	my %hash   = %{$_[0]};
    860. 

  860. 	my $method = $_[1];
    861. 

  861. 	my @methods;
    862. 

  862. 

  863. 	if (defined $method) { $methods[0] = $method; } else { @methods = (keys %hash); };
    864. 

  864. 

  865. 	my $count = 1;
    866. 

  866. 

  867. 	&Log("\tCOUNT\tMETHOD\tHITS\tIDENT\tCONSV\tSIGN\tHLENGTH\tQLENGTH\tTLENGTH\tSTART\tEND\tSTRAND\tNAME\n");
    868. 

  868. 	foreach my $method (sort @methods)
    869. 

  869. 	{
    870. 

  870. 		foreach my $name (sort keys %{$hash{$method}})
    871. 

  871. 		{
    872. 

  872. 			my $valid = 1;
    873. 

  873. 			my $value = $hash{$method}{$name};
    874. 

  874. 			my $data  = &getCharStr($method,$name);
    875. 

  875. 	
    876. 

  876. 			if ( ( ! $hypothetical ) && ( $name =~ /hypothetical.protein/sig ) && ( ! ( $name =~ /conserved/sig ) ) ) { $valid = 0; };
    877. 

  877. 	
    878. 

  878. 			if ($value == 1)
    879. 

  879. 			{
    880. 

  880. 				if ($valid) { &Log("\t" . $count++ . "\t$data\t$name\n") };
    881. 

  881. 			}
    882. 

  882. 			elsif ($value >= 2)
    883. 

  883. 			{
    884. 

  884. 				if ($valid) { &Log("\t" . $count++ . "\t$data\t$name\n") };
    885. 

  885. 			}
    886. 

  886. 			else
    887. 

  887. 			{
    888. 

  888. 				&Log("GENE $name SHOUDNT BE HERE\n");
    889. 

  889. 			}
    890. 

  890. 		}
    891. 

  891. 	}
    892. 

  892. }
    893. 

  893. 

  894. 

  895. sub getCharStr
    896. 

  896. {
    897. 

  897. 	my $method      = $_[0];
    898. 

  898. 	my $query_name  = $_[1];
    899. 

  899. 	my $result;
    900. 

  900. 	my $colums = 10;
    901. 

  901. 

  902. 	my %data = %{&getChar($method,$query_name)};
    903. 

  903. 

  904. 	if (%data)
    905. 

  905. 	{
    906. 

  906. 		my $ident = $data{"ident"};
    907. 

  907. 		my $consv = $data{"consv"};
    908. 

  908. 		$result = "$method\t" . $data{"hits"} . "\t" . $ident . "\t" . $consv . "\t" . $data{"sign"} . "\t" . $data{"Hlength"} . "\t" . $data{"Qlength"} . "\t" . $data{"Tlength"} . "\t" . $data{"start"} . "\t" . $data{"end"} . "\t" . $data{"strand"};
    909. 

  909. 	}
    910. 

  910. 	else
    911. 

  911. 	{
    912. 

  912. 		$result = "\t"x$colums;
    913. 

  913. 	}
    914. 

  914. 

  915. 	return $result;
    916. 

  916. }
    917. 

  917. 

  918. sub roundFrac
    919. 

  919. {
    920. 

  920. 	my $number = $_[0];
    921. 

  921. 

  922. 	$number = sprintf("%03.1f", $number);
    923. 

  923. 

  924. 	return $number;
    925. 

  925. }
    926. 

  926. 

  927. sub percentage
    928. 

  928. {
    929. 

  929. 	my $value = $_[0];
    930. 

  930. 	my $total = $_[1];
    931. 

  931. 	my $result;# = sprintf("%.1f", (($value/$total)*100));
    932. 

  932. 

  933. 	$value    = (($value/$total)*100);
    934. 

  934. 

  935. # 	if ($value < 10)
    936. 

  936. # 	{
    937. 

  937. # 		$result = sprintf("%.1f", $value);
    938. 

  938. 		$result = sprintf("%05.1f", $value);
    939. 

  939. # 	}
    940. 

  940. 

  941. 	return $result;
    942. 

  942. }
    943. 

  943. 

  944. 1;
    945. 

  945.