cgi-bin/methods/unigene/unigene-validation.pl

   1 use strict;
   2 use CXGN::Page;
   3 my $page=CXGN::Page->new('unigene-validation.html','html2pl converter');
   4 $page->header('Assembly Process Validation');
   5 print<<END_HEREDOC;
   6
   7   <center>
   8
   9     <table summary="" width="720" cellpadding="0" cellspacing="0"
  10     border="0">
  11       <tr>
  12         <td>
  13
  14
  15           <h3>Assembly Process Validation</h3>
  16
  17           <p>In an effort to validate SGN's unigene assembly
  18           process, we have attempted to compare our combined
  19           Lycopersicon build with <a href=
  20           "http://www.tigr.org/tdb/tgi/">TIGR's tomato gene
  21           index</a>. These comparisons are based on the latest TIGR
  22           tomato gene index available at the time, published on
  23           June 1, 2002. It is noted here that neither SGN's unigene
  24           nor TIGR's gene index builds are supported by
  25           experimental evidence, and thus both remain
  26           approximations of the true nature of the genomes
  27           represented.</p>
  28
  29           <p>Due to differences in input data, such as EST
  30           sequences not common to both builds, and differences in
  31           chromatogram processing, direct comparison of the two
  32           builds exposes mostly "noisy" differences that lead to
  33           inconclusive results in attempts to characterize or
  34           manually curate the observed differences.</p>
  35
  36           <p>Thus, the data presented below serves to indicate the
  37           observed similarity between builds and demonstrate that
  38           neither build differs significantly from the other
  39           indicating a suspicious assembly process. See <a href=
  40           "unigene-methods.pl">this page</a> for a discussion on
  41           the assembly process.</p>
  42
  43           <table summary="" border="1">
  44             <tr>
  45               <td></td>
  46               <td>SGN Lycopersicon combined build #1</td>
  47               <td>TIGR Tomato Gene Index</td>
  48             </tr>
  49
  50             <tr>
  51               <td>Total # of output sequences</td>
  52               <td>31278</td>
  53               <td>31102</td>
  54             </tr>
  55
  56             <tr>
  57               <td>Contigs (TCs)</td>
  58               <td>16200</td>
  59               <td>15211</td>
  60             </tr>
  61
  62             <tr>
  63               <td>Singlets</td>
  64               <td>15078</td>
  65               <td>15891</td>
  66             </tr>
  67
  68             <tr>
  69               <td>Censored inputs</td>
  70               <td>14310</td>
  71               <td>11054</td>
  72             </tr>
  73
  74             <tr>
  75               <td>Exclusive Contigs</td>
  76               <td>0</td>
  77               <td>0</td>
  78             </tr>
  79
  80             <tr>
  81               <td>Exclusive Singlets</td>
  82               <td>2044</td>
  83               <td>707</td>
  84             </tr>
  85           </table>
  86
  87           <p><strong>Contigs</strong> are unigenes or gene index
  88           sequences which are composed of the consensus of an
  89           alignment of two or more EST sequences.
  90           <strong>Singlets</strong> are sequences which have been
  91           determined not to overlap sufficiently with any other
  92           sequence in the input data set. <strong>Censored
  93           inputs</strong> are input sequences which are not common
  94           to both sets. <strong>Exclusive contigs</strong> are
  95           contigs composed entirely of input sequences which are
  96           not common to both builds. <strong>Exclusive
  97           singlets</strong> are singlets found only in the
  98           indicated build. Since no exclusive contigs were found,
  99           this indicates that every contig in SGN's build, and
 100           every TC in TIGR's tomato gene index is represented by at
 101           least one common input sequence for both
 102           builds.</p>
 103
 104           <p>After normalizing the unigene membership data to
 105           compare solely in terms of input sequences common to both
 106           builds, we find:</p>
 107
 108           <table summary="" border="1">
 109             <tr>
 110               <td></td>
 111               <td>SGN</td>
 112               <td>TIGR</td>
 113             </tr>
 114
 115             <tr>
 116               <td>Total # of output sequences</td>
 117               <td>29234</td>
 118               <td>30395</td>
 119             </tr>
 120
 121             <tr>
 122               <td>Contigs (TCs)</td>
 123               <td>15034</td>
 124               <td>14432</td>
 125             </tr>
 126
 127             <tr>
 128               <td>Singlets</td>
 129               <td>14200</td>
 130               <td>15963</td>
 131             </tr>
 132           </table><br />
 133
 134           <p>Since the input sequences have been normalized to a
 135           common set at this point, and output sequences which are
 136           resultant of exclusively non-common sequences are removed
 137           from consideration, this data suggests that SGN's
 138           assembly process is slightly more lenient, allowing the
 139           assembly of more sequences in to contigs. We find here
 140           that 74.5\% of SGN unigene build is identical to TIGR's
 141           gene index. Most of the remaining differences turn out to
 142           be cases where a contig in SGN is represented in TIGR as
 143           one contig and one or more singlets, or vice versa.
 144           Investigation of these cases is consistent with the claim
 145           above, that SGN's build is biased slightly toward
 146           inclusion of sequences into contigs. Although above it
 147           indicates that 2044 singlets are exclusive to SGN, the
 148           number of singlets has not dropped by 2044 becuase some
 149           contigs have become singlets after censoring non-common
 150           input sequences from consideration. The same is true for
 151           TIGR's build.</p>
 152
 153           <p>Since the Lycopersicon combined build and TIGR's
 154           tomato gene index contain data from 3 different
 155           Lycopersicon species, its useful to look at the number of
 156           unigenes specific to <em>Lycopersicon hirsutum</em> and
 157           <em>Lycopersicon pennellii</em>, which ought to show
 158           substantial allelic variation with the species dominantly
 159           represented in the input data, <em>Lycopersicon
 160           esculentum</em>.</p>
 161
 162           <table summary="" border="1">
 163             <tr>
 164               <td></td>
 165               <td>SGN</td>
 166               <td>TIGR</td>
 167             </tr>
 168
 169             <tr>
 170               <td><em>hirsutum</em> specific contigs</td>
 171               <td>94</td>
 172               <td>157</td>
 173             </tr>
 174
 175             <tr>
 176               <td><em>pennellii</em> specific contigs</td>
 177               <td>147</td>
 178               <td>113</td>
 179             </tr>
 180
 181             <tr>
 182               <td><em>hirsutum/esculentum</em> mixed contigs</td>
 183               <td>1908</td>
 184               <td>1863</td>
 185             </tr>
 186
 187             <tr>
 188               <td><em>pennellii/esculentum</em> mixed contigs</td>
 189               <td>6552</td>
 190               <td>6624</td>
 191             </tr>
 192           </table>
 193
 194           <p>From this data, both TIGR and SGN's assembly processes
 195           are allowing the contig assembly of sequences which
 196           contain small evolutionary divergence as well as
 197           sequencing errors. It is not clear from this data whether
 198           or not orthologs are specifically isolated in the
 199           assembly. Neither assembly process at this time contains
 200           specific steps for isolating orthologs from paralogs in
 201           cross-species assemblies. This question can not be
 202           completely settled <em>in silico</em>.</p>
 203
 204           <p>In conclusion, we find that the insight gained from
 205           comparing TIGR's gene index with SGN's Lycopersicon
 206           combined unigene build indicates that each procedure
 207           confirms the predictions of the other in most cases.
 208           Differences are observed, but most are attributable to
 209           differences in inputs to the processes. The reader is
 210           reminded that the above data attempts to characterize the
 211           differences in outputs of two separate processes, while
 212           <strong>not</strong> being able to control the
 213           differences in inputs. Thus, the conclusive power of the
 214           analysis is limited.</p>
 215         </td>
 216       </tr>
 217
 218     </table>
 219   </center>
 220 END_HEREDOC
 221 $page->footer();