FUSIONSEQ: showDetails_cgi.c Source File

00001 #include "log.h"
00002 #include "format.h"
00003 #include "html.h"
00004 #include "htmlLinker.h"
00005 #include "gfr.h"
00006 #include "geneFusionsConfig.h"
00007 
00008 #define GFR_PAIR_NAME_EXONIC_EXONIC "exon-exon"
00009 #define GFR_PAIR_NAME_EXONIC_INTRONIC "exon-intron"
00010 #define GFR_PAIR_NAME_EXONIC_JUNCTION "exon-boundary"
00011 #define GFR_PAIR_NAME_INTRONIC_EXONIC "intron-exon"
00012 #define GFR_PAIR_NAME_INTRONIC_INTRONIC "intron-intron"
00013 #define GFR_PAIR_NAME_INTRONIC_JUNCTION "intron-boundary"
00014 #define GFR_PAIR_NAME_JUNCTION_JUNCTION "boundary-boundary"
00015 #define GFR_PAIR_NAME_JUNCTION_EXONIC "boundary-exon"
00016 #define GFR_PAIR_NAME_JUNCTION_INTRONIC "boundary-intron"
00017 
00018 
00019 static char* processString (char *str)
00020 {
00021   static Stringa buffer = NULL;
00022   Texta tokens;
00023   int i;
00024 
00025   if (!strchr (str,'|')) {
00026     return str;
00027   }
00028   stringCreateClear (buffer,100);
00029   tokens = textStrtokP (str,"|");
00030   for (i = 0; i < arrayMax (tokens); i++) {
00031     stringAppendf (buffer,"%s%s",textItem (tokens,i),i < arrayMax (tokens) - 1 ? "<br>" : "");
00032   }
00033   return string (buffer);
00034 }  
00035 
00036 char* getPairTypeName(  int pairType ) {
00037   switch ( pairType ) {
00038   case GFR_PAIR_TYPE_EXONIC_EXONIC:
00039     return GFR_PAIR_NAME_EXONIC_EXONIC;
00040   case GFR_PAIR_TYPE_EXONIC_INTRONIC:
00041     return GFR_PAIR_NAME_EXONIC_INTRONIC;
00042   case GFR_PAIR_TYPE_EXONIC_JUNCTION:
00043     return GFR_PAIR_NAME_EXONIC_JUNCTION;
00044   case GFR_PAIR_TYPE_INTRONIC_EXONIC:
00045     return GFR_PAIR_NAME_INTRONIC_EXONIC;
00046   case GFR_PAIR_TYPE_INTRONIC_INTRONIC:
00047     return GFR_PAIR_NAME_INTRONIC_INTRONIC;
00048   case GFR_PAIR_TYPE_INTRONIC_JUNCTION:
00049     return GFR_PAIR_NAME_INTRONIC_JUNCTION;
00050   case GFR_PAIR_TYPE_JUNCTION_JUNCTION:
00051     return GFR_PAIR_NAME_JUNCTION_JUNCTION;
00052   case GFR_PAIR_TYPE_JUNCTION_EXONIC:
00053     return GFR_PAIR_NAME_JUNCTION_EXONIC;    
00054   case GFR_PAIR_TYPE_JUNCTION_INTRONIC:
00055     return GFR_PAIR_NAME_JUNCTION_INTRONIC;
00056   }
00057   return "NULL";
00058 }
00059 char* getEntryNumber( int number, int pairType, int readNum ) {
00060   int flag=0;
00061   Stringa str=stringCreate(10);
00062   switch ( pairType ) {
00063   case GFR_PAIR_TYPE_EXONIC_EXONIC:    
00064   case GFR_PAIR_TYPE_EXONIC_INTRONIC:
00065   case GFR_PAIR_TYPE_INTRONIC_EXONIC:   
00066   case GFR_PAIR_TYPE_INTRONIC_INTRONIC:
00067     break;
00068   case GFR_PAIR_TYPE_INTRONIC_JUNCTION:
00069   case GFR_PAIR_TYPE_EXONIC_JUNCTION:
00070     if( readNum==2 ) flag=1;      
00071     break;
00072   case GFR_PAIR_TYPE_JUNCTION_EXONIC:
00073   case GFR_PAIR_TYPE_JUNCTION_INTRONIC:
00074     if( readNum==1 ) flag=1;
00075     break;
00076   case GFR_PAIR_TYPE_JUNCTION_JUNCTION:
00077     flag=1;
00078   }
00079   if( flag==1 ) {
00080     int rem = number % 2;
00081     if( rem == 0 )
00082       stringPrintf(str, "%d right", number/2);
00083     else 
00084       stringPrintf(str, "left %d", number/2+1);
00085   } else {
00086     stringPrintf(str, "%d", number);
00087   }
00088   return string(str);
00089 }
00090 
00091 int main (int argc, char *argv[]) 
00092 {
00093   FILE* ftmp=NULL;
00094   cgiInit();
00095   cgiHeader("text/html");
00096 
00097   if (argc == 3) {
00098     GfrEntry *currGE;
00099     Stringa buffer;
00100     GfrPairCount *currGEPC;
00101     GfrInterRead *currGIR;
00102     int i;
00103 
00104     puts ("<html>");
00105     puts ("<head>");
00106     html_printGenericStyleSheet (12);
00107     puts ("<title>geneFusions Details</title>\n");
00108     puts ("</head>");
00109     puts ("<body>");
00110     buffer = stringCreate (100);
00111     stringPrintf (buffer,"%s/%s.gfr",WEB_DATA_DIR,argv[1]);    
00112     gfr_init (string (buffer));
00113     while (currGE = gfr_nextEntry ()){
00114       fflush( stdout );
00115       if (!strEqual (currGE->id,argv[2])) {
00116         continue;
00117       }
00118       printf ("<h1>Detailed summary for potential gene fusion candidate</h1><br>");
00119       puts ("<table border=0 cellpadding=10>");
00120       puts ("<tr align=left valign=top>");
00121       puts ("<td width=400>");
00122       puts ("<h2>Summary information</h2><br>");
00123       printf ("<b>Identifier</b>: %s<br><br>\n",currGE->id);
00124       printf ("<b>Number of inter paired-end reads</b>: %d<br><br>\n",currGE->numInter);
00125       printf ("<b>Type</b>: %s<br><br>\n",currGE->fusionType);     
00126       
00127       stringPrintf(buffer, "%s/GFF/%s.gff", WEB_DATA_DIR,currGE->id);       
00128       ftmp = fopen( string(buffer), "r" ); // displaying this only if data are present
00129       if( ftmp ) {
00130          printf ("<b>Connected Reads</b>: <a href=%s&hgt.customText=%s/GFF/%s.gff target=blank>UCSC connectivity graph</a><br>\n",
00131               htmlLinker_generateLinkToGenomeBrowserAtUCSC ("hg18","vertebrate","human",currGE->chromosomeTranscript1,currGE->startTranscript1 - UCSC_GENOME_BROWSER_FLANKING_REGION,currGE->endTranscript2 + UCSC_GENOME_BROWSER_FLANKING_REGION),
00132               WEB_DATA_LINK,currGE->id); 
00133          fclose( ftmp );
00134       }
00135    
00136       puts ("</td>");
00137       puts ("<td>");
00138       puts ("<h2>Transcript connectivity graph</h2>");
00139       printf ("<img src=%s/IMAGES/%s.jpg alt=geneFusionImage>\n",WEB_DATA_LINK,currGE->id);
00140       puts ("</td>");
00141       puts ("<td>");
00142       puts ("<h2>Transcript connectivity table</h2><br>");
00143       puts ("<table border=0>");
00144       puts ("<tr align=left>");
00145       puts ("<th width=200>Pair Type</th>");
00146       puts ("<th width=200>Entry transcript 1</th>");
00147       puts ("<th width=200>Entry transcript 2</th>");
00148       puts ("<th width=200>Counts</th>");
00149       puts ("</tr>");
00150       fflush( stdout );
00151       for (i = 0; i < arrayMax (currGE->pairCounts); i++) {
00152         currGEPC = arrp (currGE->pairCounts,i,GfrPairCount);    
00153         printf ("<tr><td>%s</td><td>%s</td><td>%s</td><td>%d</td></tr>\n", 
00154                 getPairTypeName(currGEPC->pairType), 
00155                 getEntryNumber(currGEPC->number1, currGEPC->pairType, 1),
00156                 getEntryNumber(currGEPC->number2, currGEPC->pairType, 2),
00157                 currGEPC->count);
00158       }
00159       puts ("</table>");
00160       puts ("</td>");
00161       puts ("</tr>");
00162       puts ("</table>");
00163       puts ("<br>");
00164 
00165       puts ("<h2>Transcript information</h2><br>");
00166       puts ("<table border=1 cellpadding=10 width=\"80%\">");
00167       puts ("<tr align=left>");
00168       puts ("<th width=\"20%\"></th>");
00169       puts ("<th><font color='blue'>Transcript 1</font></th>");
00170       puts ("<th><font color='orange'>Transcript 2</font></th>");
00171       puts ("</tr>");
00172       puts ("<tr align=left>");
00173       puts ("<td width=\"20%\"><b>Gene symbol(s)</b></td>");
00174       printf ("<td width=\"30%%\"><font color='blue'>%s</font></td>\n",processString (currGE->geneSymbolTranscript1));
00175       printf ("<td width=\"30%%\"><font color='orange'>%s</font></td>\n",processString (currGE->geneSymbolTranscript2));
00176       puts ("</tr>");
00177       puts ("<tr align=left>");
00178       puts ("<td width=\"20%\"><b>Coordinates</b></td>");
00179       printf ("<td width=\"30%%\">%s:%d-%d</td>\n",currGE->chromosomeTranscript1,currGE->startTranscript1,currGE->endTranscript1);
00180       printf ("<td width=\"30%%\">%s:%d-%d</td>\n",currGE->chromosomeTranscript2,currGE->startTranscript2,currGE->endTranscript2);
00181       puts ("</tr>");
00182       puts ("<tr align=left>");
00183       puts ("<td width=\"20%\"><b>Strand</b></td>");
00184       printf ("<td width=\"30%%\">%c</td>\n",currGE->strandTranscript1);
00185       printf ("<td width=\"30%%\">%c</td>\n",currGE->strandTranscript2);
00186       puts ("</tr>");
00187       puts ("<tr align=left>");
00188       puts ("<td width=\"20%\"><b>Gene description(s)</b></td>");
00189       printf ("<td width=\"30%%\">%s</td>\n",processString (currGE->descriptionTranscript1));
00190       printf ("<td width=\"30%%\">%s</td>\n",processString (currGE->descriptionTranscript2));
00191       puts ("</tr>");
00192       puts ("<tr align=left>");
00193       puts ("<td width=\"20%\"><b>Number of exons</b></td>");
00194       printf ("<td width=\"30%%\">%d</td>\n",currGE->numExonsTranscript1);
00195       printf ("<td width=\"30%%\">%d</td>\n",currGE->numExonsTranscript2);
00196       puts ("</tr>");
00197       puts ("<tr align=left>");
00198       puts ("<td width=\"20%\"><b>Number of intra paired-end reads</b></td>");
00199       printf ("<td width=\"30%%\">%d</td>\n",currGE->numIntra1);
00200       printf ("<td width=\"30%%\">%d</td>\n",currGE->numIntra2);
00201       puts ("</tr>");
00202       puts ("<tr align=left>");
00203       puts ("<td width=\"20%\"><b>Links</b></td>");
00204       printf ("<td width=\"30%%\">[<a href=%s&hgt.customText=%s/BED/%s_1.bed target=blank>UCSC genome browser</a>]&nbsp;&nbsp;&nbsp;[<a href=%s/FASTA/%s_1.fasta>FASTA file</a>]<br></td>\n",
00205               htmlLinker_generateLinkToGenomeBrowserAtUCSC ("hg18","vertebrate","human",currGE->chromosomeTranscript1,currGE->startTranscript1 - UCSC_GENOME_BROWSER_FLANKING_REGION,currGE->endTranscript1 + UCSC_GENOME_BROWSER_FLANKING_REGION),
00206               WEB_DATA_LINK,currGE->id,WEB_DATA_LINK,currGE->id); 
00207       printf ("<td width=\"30%%\">[<a href=%s&hgt.customText=%s/BED/%s_2.bed target=blank>UCSC genome browser</a>]&nbsp;&nbsp;&nbsp;[<a href=%s/FASTA/%s_2.fasta>FASTA file</a>]<br></td></tr>\n",
00208               htmlLinker_generateLinkToGenomeBrowserAtUCSC ("hg18","vertebrate","human",currGE->chromosomeTranscript2,currGE->startTranscript2 - UCSC_GENOME_BROWSER_FLANKING_REGION,currGE->endTranscript2 + UCSC_GENOME_BROWSER_FLANKING_REGION),
00209               WEB_DATA_LINK,currGE->id,WEB_DATA_LINK,currGE->id); 
00210       
00211       puts ("<tr align=left>");
00212       puts ("<td width=\"20%\"><b>Expression</b></td>"); 
00213 
00214       stringPrintf(buffer, "%s/BGRS/%s_%s.bgr.gz", WEB_DATA_DIR,argv[1],currGE->chromosomeTranscript1);  
00215       ftmp = fopen( string(buffer), "r" ); // displaying this only if data are present
00216       puts("<td width=\"30%\">");
00217       if( ftmp ) {
00218         printf ("[<a href=%s&hgt.customText=%s/BGRS/%s_%s.bgr.gz target=blank>Expression %s</a>]",
00219                 htmlLinker_generateLinkToGenomeBrowserAtUCSC ("hg18","vertebrate","human",currGE->chromosomeTranscript1,currGE->startTranscript1 - UCSC_GENOME_BROWSER_FLANKING_REGION,currGE->endTranscript1 + UCSC_GENOME_BROWSER_FLANKING_REGION),
00220                 WEB_DATA_LINK,argv[1],currGE->chromosomeTranscript1,currGE->chromosomeTranscript1); 
00221         fclose(ftmp);
00222       }
00223       puts("</td>");
00224 
00225       stringPrintf(buffer, "%s/BGRS/%s_%s.bgr.gz", WEB_DATA_DIR,argv[1],currGE->chromosomeTranscript2); 
00226       ftmp = fopen( string(buffer), "r" ); // displaying this only if data are present
00227       puts("<td width=\"30%\">");
00228       if( ftmp ) {
00229         printf ("[<a href=%s&hgt.customText=%s/BGRS/%s_%s.bgr.gz target=blank>Expression %s</a>]",
00230                 htmlLinker_generateLinkToGenomeBrowserAtUCSC ("hg18","vertebrate","human",currGE->chromosomeTranscript2,currGE->startTranscript2 - UCSC_GENOME_BROWSER_FLANKING_REGION,currGE->endTranscript2 + UCSC_GENOME_BROWSER_FLANKING_REGION),
00231                 WEB_DATA_LINK,argv[1],currGE->chromosomeTranscript2,currGE->chromosomeTranscript2); 
00232         fclose(ftmp);
00233       } 
00234       puts("</td>");
00235       puts("</tr>");
00236       puts ("</table><br><br>");
00237       
00238       puts ("<h2>Breakpoint analysis</h2><br>");
00239       puts ("<table border=1 width=\"80%\" cellpadding=10><thead><tr><th>Orientation</th><th>Alignments</th><th colspan=2>Breakpoints</th></tr></thead><tbody>");
00240       puts ("<tr><td>Orientation AB</td>");
00241         if (currGE->strandTranscript1=='+') {
00242           currGE->strandTranscript2=='+' ? stringPrintf(buffer, "AB_trans1F_trans2F") : stringPrintf(buffer, "AB_trans1F_trans2R");
00243         } else if( currGE->strandTranscript1 == '-') {
00244           currGE->strandTranscript2=='+' ? stringPrintf(buffer, "AB_trans1R_trans2F") : stringPrintf(buffer, "AB_trans1R_trans2R");
00245         } else {
00246           die("Strand informatation is not correct (transcript 1): %c", currGE->strandTranscript1);
00247         }
00248         printf ("<td align=center><a href=%s/ALIGNMENTS/%s_AB_breakPointAlignments.txt><img src=%s/IMAGES/%s.png></img>&nbsp;AB</a></td>",WEB_DATA_LINK,currGE->id, WEB_DATA_LINK, string(buffer)); 
00249         printf ("<td align=center><a href=%s&hgt.customText=%s/WIGS/%s_AB_breakPointsTranscript1.wig target=blank>Breakpoints transcript 1 UCSC Genome Browser</a></td>", htmlLinker_generateLinkToGenomeBrowserAtUCSC ("hg18","vertebrate","human",currGE->chromosomeTranscript1,currGE->startTranscript1 - UCSC_GENOME_BROWSER_FLANKING_REGION,currGE->endTranscript1 + UCSC_GENOME_BROWSER_FLANKING_REGION),WEB_DATA_LINK,currGE->id);
00250         printf ("<td align=center><a href=%s&hgt.customText=%s/WIGS/%s_AB_breakPointsTranscript2.wig target=blank>Breakpoints transcript 2 UCSC Genome Browser</a></td></tr>", htmlLinker_generateLinkToGenomeBrowserAtUCSC ("hg18","vertebrate","human",currGE->chromosomeTranscript2,currGE->startTranscript2 - UCSC_GENOME_BROWSER_FLANKING_REGION,currGE->endTranscript2 + UCSC_GENOME_BROWSER_FLANKING_REGION),WEB_DATA_LINK,currGE->id);  
00251       fflush(stdout);
00252       puts   ("<tr><td>Orientation BA</td>");  
00253       if (currGE->strandTranscript1 == '+') {
00254         currGE->strandTranscript2=='+' ? stringPrintf(buffer, "BA_trans1F_trans2F") : stringPrintf(buffer, "BA_trans1F_trans2R");
00255       } else if( currGE->strandTranscript1 == '-') {
00256         currGE->strandTranscript2=='+' ? stringPrintf(buffer, "BA_trans1R_trans2F") : stringPrintf(buffer, "BA_trans1R_trans2R");
00257       } else {
00258         die("Strand informatation is not correct (transcript2): %c", currGE->strandTranscript2);
00259         }       
00260       printf ("<td align=center><a href=%s/ALIGNMENTS/%s_BA_breakPointAlignments.txt><img src=%s/IMAGES/%s.png></img>&nbsp;BA</a></td>",WEB_DATA_LINK,currGE->id, WEB_DATA_LINK,string(buffer));
00261       printf ("<td align=center><a href=%s&hgt.customText=%s/WIGS/%s_BA_breakPointsTranscript2.wig target=blank>Breakpoints transcript 2 UCSC Genome Browser</a></td>", htmlLinker_generateLinkToGenomeBrowserAtUCSC ("hg18","vertebrate","human",currGE->chromosomeTranscript2,currGE->startTranscript2 - UCSC_GENOME_BROWSER_FLANKING_REGION,currGE->endTranscript2 + UCSC_GENOME_BROWSER_FLANKING_REGION),WEB_DATA_LINK,currGE->id);
00262       printf ("<td align=center><a href=%s&hgt.customText=%s/WIGS/%s_BA_breakPointsTranscript1.wig target=blank>Breakpoints transcript 1 UCSC Genome Browser</a></td></tr>", htmlLinker_generateLinkToGenomeBrowserAtUCSC ("hg18","vertebrate","human",currGE->chromosomeTranscript1,currGE->startTranscript1 - UCSC_GENOME_BROWSER_FLANKING_REGION,currGE->endTranscript1 + UCSC_GENOME_BROWSER_FLANKING_REGION),WEB_DATA_LINK,currGE->id);       
00263 
00264       puts ("</tbody></table>");
00265       puts ("<br><br><br>");
00266       fflush(stdout);
00267     
00268     
00269       puts ("<h2>Read coordinates</h2><br>");
00270       puts ("<table border=0>");
00271       puts ("<tr align=left>");
00272       puts ("<th width=\"10%\">Pair Type</th>");
00273       puts ("<th width=\"10%\">Entry Transcript 1</th>");
00274       puts ("<th width=\"10%\">Read start transcript 1</th>");
00275       puts ("<th width=\"10%\">Read end transcript 1</th>");
00276       puts ("<th width=\"10%\">Entry Transcript 2</th>");
00277       puts ("<th width=\"10%\">Read start transcript 2</th>");
00278       puts ("<th width=\"10%\">Read end transcript 2</th>");
00279       puts ("</tr>");     
00280       for (i = 0; i < arrayMax (currGE->interReads); i++) {
00281         currGIR = arrp (currGE->interReads,i,GfrInterRead);
00282         printf ("<tr><td>%s</td><td>%s</td><td>%d</td><td>%d</td><td>%s</td><td>%d</td><td>%d</td></tr>\n",
00283                 getPairTypeName(currGIR->pairType), 
00284                 getEntryNumber(currGIR->number1, currGIR->pairType, 1),
00285                 currGIR->readStart1,currGIR->readEnd1,
00286                 getEntryNumber(currGIR->number2,currGIR->pairType, 2),
00287                 currGIR->readStart2,
00288                 currGIR->readEnd2);
00289       }
00290       puts ("</table><br><br><br>");
00291       puts ("</body>");
00292       puts ("</html>");
00293     fflush (stdout);
00294     }
00295   }
00296   return 0;
00297 }