int main(int argc, char *argv[])
/* Process command line. */
{
optionInit(&argc, argv, options);
if (argc != 4)
usage();
database=argv[1];
hSetDb(argv[1]);
protDbName = hPdbFromGdb(database);
fbName(argv[1], protDbName, argv[2], argv[3]);
return 0;
}
void doMiddle(struct cart *theCart)
/* Print the body of an html file. */
{
char cond_str[255];
struct sqlConnection *conn;
char *proteinAC;
char *chp, *chp1, *chp9;
char *debugTmp = NULL;
char *chromStr, *cdsStartStr, *cdsEndStr, posStr[255];
char *supportedGenomeDatabase;
char *answer;
char *queryID;
/* Initialize layout and database. */
cart = theCart;
/* Uncomment this to see parameters for debugging. */
/* Be careful though, it breaks if custom track
* is more than 4k */
/*
{ struct dyString *state = cgiUrlString();
hPrintf("State: %s\n", state->string);
}
*/
queryID = cartOptionalString(cart, "proteinID");
if (sameString(queryID, ""))
{
hUserAbort("Please go back and enter a gene symbol or a Swiss-Prot/TrEMBL protein ID.\n");
}
if (cgiVarExists("db"))
{
/* if db is known, get key variables set */
proteinInSupportedGenome = TRUE;
database = cgiOptionalString("db");
organism = hDbOrganism(database);
protDbName = hPdbFromGdb(database);
proteinID = strdup(queryID);
}
else
{
protCntInSwissByGene = searchProteinsInSwissProtByGene(queryID);
/* no CGI 'db' variable means it did not come in from GB but from pbGateway */
/* search existing GB databases to see if this protein can be found */
protCntInSupportedGenomeDb =
searchProteinsInSupportedGenomes(queryID, &supportedGenomeDatabase);
if ((protCntInSupportedGenomeDb > 1) || protCntInSwissByGene >= 1)
{
/* more than 1 proteins match the query ID, present selection web page */
proteinInSupportedGenome = 1;
presentProteinSelections(queryID, protCntInSwissByGene, protCntInSupportedGenomeDb);
return;
}
else
{
if (protCntInSupportedGenomeDb == 1)
{
/* one and only one protein found in a genome DB that support KG and PB */
proteinInSupportedGenome = TRUE;
database = strdup(supportedGenomeDatabase);
organism = hDbOrganism(database);
protDbName = hPdbFromGdb(database);
proteinID=strdup(queryID);
}
else
{
/* not found in genome DBs that support KG/PB */
/* now search PROTEOME_DB_NAMES to see if this protein is there. */
answer = uniProtFindPrimAcc(queryID);
if (answer == NULL)
{
hUserAbort("'%s' does not seem to be a valid UniProtKB protein ID or a gene "
"symbol.<br><br>Click <A HREF=\"../cgi-bin/pbGateway\">here</A> "
"to start another query.", queryID);
}
proteinInSupportedGenome = FALSE;
database = strdup(GLOBAL_PB_DB);
organism = strdup("");
protDbName = strdup(PROTEOME_DB_NAME);
proteinID = strdup(answer);
}
}
if (proteinInSupportedGenome)
{
spConn = sqlConnect(database);
sqlSafefFrag(cond_str, sizeof(cond_str), "alias='%s'", queryID);
proteinID = sqlGetField(database, "kgSpAlias", "spID", cond_str);
sqlSafefFrag(cond_str, sizeof(cond_str), "spID='%s'", proteinID);
answer = sqlGetField(database, "kgXref", "spDisplayID", cond_str);
sqlSafefFrag(cond_str, sizeof(cond_str), "proteinID='%s'", answer);
chromStr = sqlGetField(database, "knownGene", "chrom", cond_str);
if (chromStr)
{
cdsStartStr = sqlGetField(database, "knownGene", "cdsStart", cond_str);
cdsEndStr = sqlGetField( database, "knownGene", "cdsEnd", cond_str);
safef(posStr, sizeof(posStr), "%s:%s-%s", chromStr, cdsStartStr, cdsEndStr);
positionStr = strdup(posStr);
cartSetString(cart, "position", positionStr);
cartSetString(cart, "organism", organism);
}
}
}
/* print out key variables for debugging */
/* printf("<br>before enter main section: <br>proteinInSupportedGenome=%d<br>proteinID=%s <br>database=%s <br>organism=%s <br>protDbName=%s\n",
proteinInSupportedGenome, proteinID, database, organism, protDbName);fflush(stdout);
*/
if (hTableExists(database, "kgProtMap2"))
{
kgVersion = KG_III;
strcpy(kgProtMapTableName, "kgProtMap2");
}
debugTmp = cartUsualString(cart, "hgDebug", "off");
if(sameString(debugTmp, "on"))
hgDebug = TRUE;
else
hgDebug = FALSE;
conn = hAllocConn(database);
hgsid = cartOptionalString(cart, "hgsid");
if (hgsid != NULL)
{
safef(hgsidStr, sizeof(hgsidStr), "&hgsid=%s", hgsid);
}
else
{
strcpy(hgsidStr, "");
}
/* check proteinID to see if it is a valid SWISS-PROT/TrEMBL accession or display ID */
/* then assign the accession number to global variable proteinID */
sqlSafefFrag(cond_str, sizeof(cond_str), "accession='%s'", proteinID);
proteinAC = sqlGetField(protDbName, "spXref3", "accession", cond_str);
if (proteinAC == NULL)
{
sqlSafefFrag(cond_str, sizeof(cond_str), "displayID='%s'", proteinID);
proteinAC = sqlGetField(protDbName, "spXref3", "accession", cond_str);
if (proteinAC == NULL)
{
hUserAbort("'%s' does not seem to be a valid Swiss-Prot/TrEMBL protein ID or gene symbol.<br><br>Click <A HREF=\"../cgi-bin/pbGateway\">here</A> to start another query."
, proteinID);
}
else
{
protDisplayID = proteinID;
proteinID = proteinAC;
}
}
else
{
sqlSafefFrag(cond_str, sizeof(cond_str), "accession='%s'", proteinID);
protDisplayID = sqlGetField(protDbName, "spXref3", "displayID", cond_str);
}
if (proteinInSupportedGenome)
{
if (kgVersion == KG_III)
{
sqlSafefFrag(cond_str, sizeof(cond_str), "spId='%s'", proteinID);
mrnaID = sqlGetField(database, "kgXref", "kgId", cond_str);
}
else
{
sqlSafefFrag(cond_str, sizeof(cond_str), "proteinID='%s'", protDisplayID);
mrnaID = sqlGetField(database, "knownGene", "name", cond_str);
}
}
else
{
mrnaID = NULL;
positionStr = NULL;
}
sqlSafefFrag(cond_str, sizeof(cond_str), "accession='%s'", proteinID);
description = sqlGetField(protDbName, "spXref3", "description", cond_str);
if (positionStr != NULL)
{
chp = strstr(positionStr, ":");
*chp = '\0';
prevGBChrom = cloneString(positionStr);
chp1 = chp + 1;
chp9 = strstr(chp1, "-");
*chp9 = '\0';
prevGBStartPos = atoi(chp1);
chp1 = chp9 + 1;
prevGBEndPos = atoi(chp1);
}
else
{
prevGBChrom = NULL;
prevGBStartPos = -1;
prevGBEndPos = -1;
}
/* Do main display. */
if (cgiVarExists("pbt.psOutput"))
handlePostscript();
else
{
doTrackForm(NULL, NULL);
}
}
static void domainsPrint(struct section *section,
struct sqlConnection *conn, char *geneId)
/* Print out protein domains. */
{
char *db = sqlGetDatabase(conn);
struct slName *el, *list;
list = spExtDbAcc1List(spConn, swissProtAcc, "Interpro");
if (list != NULL)
{
char query[256], **row, **row2;
struct sqlResult *sr, *sr2;
hPrintf("<B>InterPro Domains: </B> ");
hPrintf("<A HREF=\"http://www.ebi.ac.uk/interpro/protein/%s\" TARGET=_blank>",
swissProtAcc);
hPrintf("Graphical view of domain structure</A><BR>");
sqlSafef(query, sizeof(query),
"select extAcc1,extAcc2 from extDbRef,extDb"
" where extDbRef.acc = '%s'"
" and extDb.val = 'Interpro' and extDb.id = extDbRef.extDb"
, swissProtAcc);
sr = sqlGetResult(spConn, query);
while ((row = sqlNextRow(sr)) != NULL)
{
//hPrintf("<A HREF=\"http://www.ebi.ac.uk/interpro/entry/%s\" TARGET=_blank>", row[0]);
//hPrintf("%s</A> - %s<BR>\n", row[0], row[1]);
char interPro[256];
char *pdb = hPdbFromGdb(db);
safef(interPro, 128, "%s.interProXref", pdb);
if (hTableExists(db, interPro))
{
sqlSafef(query, sizeof(query),
"select description from %s where accession = '%s' and interProId = '%s'",
interPro, swissProtAcc, row[0]);
sr2 = sqlGetResult(conn, query);
if ((row2 = sqlNextRow(sr2)) != NULL)
{
hPrintf("<A HREF=\"http://www.ebi.ac.uk/interpro/entry/%s\" TARGET=_blank>", row[0]);
hPrintf("%s</A> - %s <BR>\n", row[0], row2[0]);
}
sqlFreeResult(&sr2);
}
else
{
hPrintf("<A HREF=\"http://www.ebi.ac.uk/interpro/entry/%s\" TARGET=_blank>", row[0]);
hPrintf("%s</A> - %s<BR>\n", row[0], row[1]);
}
}
hPrintf("<BR>\n");
slFreeList(&list);
}
if (kgVersion == KG_III)
{
/* Do Pfam domains here. */
list = getPfamDomainList(conn, geneId);
if (list != NULL)
{
hPrintf("<B>Pfam Domains:</B><BR>");
for (el = list; el != NULL; el = el->next)
{
char query[256];
char *description;
sqlSafef(query, sizeof(query),
"select description from pfamDesc where pfamAC='%s'", el->name);
description = sqlQuickString(conn, query);
if (description == NULL)
description = cloneString("n/a");
hPrintf("<A HREF=\"http://pfam.xfam.org/family?acc=%s\" TARGET=_blank>",
el->name);
hPrintf("%s</A> - %s<BR>\n", el->name, description);
freez(&description);
}
slFreeList(&list);
hPrintf("<BR>\n");
}
/* Do SCOP domains here */
list = getDomainList(conn, geneId, "Scop");
if (list != NULL)
{
hPrintf("<B>SCOP Domains:</B><BR>");
for (el = list; el != NULL; el = el->next)
{
char query[256];
char *description;
sqlSafef(query, sizeof(query),
"select description from scopDesc where acc='%s'", el->name);
description = sqlQuickString(conn, query);
if (description == NULL)
description = cloneString("n/a");
hPrintf("<A HREF=\"http://scop.berkeley.edu/sunid=%s\" TARGET=_blank>",
el->name);
hPrintf("%s</A> - %s<BR>\n", el->name, description);
freez(&description);
}
slFreeList(&list);
hPrintf("<BR>\n");
}
}
else
{
list = spExtDbAcc1List(spConn, swissProtAcc, "Pfam");
if (list != NULL)
{
char *pfamDescSql = genomeSetting("pfamDescSql");
hPrintf("<B>Pfam Domains:</B><BR>");
for (el = list; el != NULL; el = el->next)
{
char query[256];
char *description;
sqlSafef(query, sizeof(query), pfamDescSql, el->name);
description = sqlQuickString(conn, query);
if (description == NULL)
description = cloneString("n/a");
hPrintf("<A HREF=\"http://pfam.xfam.org/family?acc=%s\" TARGET=_blank>",
el->name);
hPrintf("%s</A> - %s<BR>\n", el->name, description);
freez(&description);
}
slFreeList(&list);
hPrintf("<BR>\n");
}
}
list = spExtDbAcc1List(spConn, swissProtAcc, "PDB");
if (list != NULL)
{
struct sqlConnection *conn2 = sqlConnect(db);
char query[256], **row;
struct sqlResult *sr;
int column = 0, maxColumn=3, rowCount=0;
hPrintf("<B>Protein Data Bank (PDB) 3-D Structure</B><BR>");
sqlSafef(query, sizeof(query),
"select extAcc1,extAcc2 from extDbRef,extDb"
" where extDbRef.acc = '%s'"
" and extDb.val = 'PDB' and extDb.id = extDbRef.extDb"
, swissProtAcc);
sr = sqlGetResult(spConn, query);
hPrintf("<TABLE><TR>\n");
while ((row = sqlNextRow(sr)) != NULL)
{
if (++column > maxColumn)
{
hPrintf("</TR><TR>");
column = 1;
if (rowCount == 0)
{
hPrintf("<TD ALIGN=CENTER COLSPAN=4><I>To conserve bandwidth, only the images from the first %d structures are shown.</I>", maxColumn);
hPrintf("</TR><TR>");
}
++rowCount;
}
hPrintf("<TD>");
hPrintf("<A HREF=\"http://www.rcsb.org/pdb/cgi/explore.cgi?pdbId=%s\" TARGET=_blank>", row[0]);
if (rowCount < 1)
hPrintf("<IMG SRC=\"http://www.rcsb.org/pdb/images/%s_asym_r_250.jpg\"><BR>", row[0]);
hPrintf("%s</A> - %s ", row[0], row[1]);
// include links LS-SNP and to launch viewer in PDB chimera
struct tempName chimerax;
lsSnpPdbChimeraSnpAnn(conn, row[0], NULL, &chimerax);
hPrintf(" <A HREF=\"%s\">Chimera</A>", chimerax.forHtml);
if (lsSnpPdbHasPdb(conn2, row[0]))
hPrintf(" <A HREF=\"%s\" TARGET=_blank>LS-SNP</A>", lsSnpPdbGetUrlPdbSnp(row[0], NULL));
hPrintf("</TD>\n");
}
hPrintf("</TR></TABLE>\n");
hPrintf("<A href=\"../goldenPath/help/chimera.html\" TARGET=_blank>Chimera help</A>\n");
hPrintf("<BR><BR>\n");
slFreeList(&list);
sqlDisconnect(&conn2);
}
/* Do modBase link. */
{
hPrintf("<B>ModBase Predicted Comparative 3D Structure on ");
modBaseAnchor(swissProtAcc);
hPrintf("%s", swissProtAcc);
hPrintf("</A></B><BR>\n");
hPrintf("<TABLE><TR>");
hPrintf("<TD>");
modBaseAnchor(swissProtAcc);
hPrintf("\n<IMG SRC=\"https://modbase.compbio.ucsf.edu/modbase-cgi/image/modbase.jpg?database_id=%s\"></A></TD>", swissProtAcc);
hPrintf("<TD>");
modBaseAnchor(swissProtAcc);
hPrintf("\n<IMG SRC=\"https://modbase.compbio.ucsf.edu/modbase-cgi/image/modbase.jpg?database_id=%s&axis=x°ree=90\"></A></TD>", swissProtAcc);
hPrintf("<TD>");
modBaseAnchor(swissProtAcc);
hPrintf("\n<IMG SRC=\"https://modbase.compbio.ucsf.edu/modbase-cgi/image/modbase.jpg?database_id=%s&axis=y°ree=90\"></A></TD>", swissProtAcc);
hPrintf("</TR><TR>\n");
hPrintf("<TD ALIGN=CENTER>Front</TD>");
hPrintf("<TD ALIGN=CENTER>Top</TD>");
hPrintf("<TD ALIGN=CENTER>Side</TD>");
hPrintf("</TR></TABLE>\n");
hPrintf("<I>The pictures above may be empty if there is no "
"ModBase structure for the protein. The ModBase structure "
"frequently covers just a fragment of the protein. You may "
"be asked to log onto ModBase the first time you click on the "
"pictures. It is simplest after logging in to just click on "
"the picture again to get to the specific info on that model.</I>");
}
}
void presentProteinSelections(char *queryID, int protCntInSwissByGene, int protCntInSupportedGenomeDb)
/* Fuction to present a web page with proteins of different organisms */
{
char *gDatabase;
char *org = NULL;
char *spID, *displayID, *desc;
char cond_str[255];
struct sqlConnection *conn, *conn3;
char query[256], query3[512];
struct sqlResult *sr, *sr3;
char **row, **row3;
struct sqlConnection *connCentral, *proteinsConn;
char queryCentral[256];
struct sqlResult *srCentral;
char *answer;
char *taxonId, *protAcc, *protDisp, *protOrg, *protDesc;
char *oldOrg, *orgSciName;
char *pbOrgSciName[MAX_PB_ORG];
boolean pbOrgPresented[MAX_PB_ORG];
boolean skipIt;
int i, maxPbOrg;
int otherCnt;
connCentral = hConnectCentral();
hPrintf("<TABLE WIDTH=\"100%%\" BGCOLOR=\"#"HG_COL_HOTLINKS"\" BORDER=\"0\" CELLSPACING=\"0\"");
hPrintf("CELLPADDING=\"2\"><TR>\n");
hPrintf("<TD ALIGN=LEFT><A HREF=\"../index.html\">%s</A></TD>", wrapWhiteFont("Home"));
hPrintf("<TD style='text-align:center; color:#FFFFFF; font-size:medium;'>%s</TD>",
"UCSC Proteome Browser");
if (proteinInSupportedGenome)
{
hPrintf("<TD ALIGN=Right><A HREF=\"../goldenPath/help/pbTracksHelpFiles/pbTracksHelp.shtml\"");
}
else
{
hPrintf("<TD ALIGN=Right><A HREF=\"../goldenPath/help/pbTracksHelpFiles/pbTracksHelp.shtml\"");
}
hPrintf("TARGET=_blank>%s</A></TD>", wrapWhiteFont("Help"));
hPrintf("</TR></TABLE>");
hPrintf("<FONT SIZE=4><BR><B>Please select one of the following proteins:<BR><BR></B></FONT>\n");
/* remmember a list of scientific names for the genomes that supports PB */
sqlSafef(queryCentral, sizeof(queryCentral),
"select distinct dbDb.scientificName from dbDb where hgPbOk=1");
srCentral = sqlMustGetResult(connCentral, queryCentral);
row3 = sqlNextRow(srCentral);
i=0;
while (row3 != NULL)
{
pbOrgSciName[i] = strdup(row3[0]);
pbOrgPresented[i] = FALSE;
i++;
row3 = sqlNextRow(srCentral);
}
maxPbOrg = i;
/* go through each genome DB that supports PB */
sqlSafef(queryCentral, sizeof(queryCentral),
"select defaultDb.name, dbDb.organism, dbDb.scientificName from dbDb,defaultDb where hgPbOk=1 and defaultDb.name=dbDb.name");
srCentral = sqlMustGetResult(connCentral, queryCentral);
row3 = sqlNextRow(srCentral);
while (row3 != NULL)
{
gDatabase = row3[0];
org = row3[1];
orgSciName= row3[2];
protDbName = hPdbFromGdb(gDatabase);
proteinsConn = sqlConnect(protDbName);
conn = sqlConnect(gDatabase);
sqlSafefFrag(cond_str, sizeof(cond_str), "alias='%s' and spID != ''", queryID);
answer = sqlGetField(gDatabase, "kgSpAlias", "count(distinct spID)", cond_str);
if ((answer != NULL) && (!sameWord(answer, "0")))
{
/* display organism name */
hPrintf("<FONT SIZE=4><B>");
hPrintf("<A href=\"http://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?mode=Undef&name=%s&lvl=0&srchmode=1\" TARGET=_blank>%s</A>",
cgiEncode(orgSciName), orgSciName);
hPrintf(" (%s):</B></FONT>\n", org);
hPrintf("<UL>");
sqlSafef(query, sizeof(query),
"select distinct spID from %s.kgSpAlias where alias='%s' "
"and spID != ''",
gDatabase, queryID);
sr = sqlMustGetResult(conn, query);
row = sqlNextRow(sr);
while (row != NULL)
{
spID = row[0];
sqlSafefFrag(cond_str, sizeof(cond_str), "accession='%s'", spID);
displayID = sqlGetField(protDbName, "spXref3", "displayID", cond_str);
sqlSafefFrag(cond_str, sizeof(cond_str), "accession='%s'", spID);
desc = sqlGetField(protDbName, "spXref3", "description", cond_str);
/* display a protein */
hPrintf(
"<LI><A HREF=\"../cgi-bin/pbGlobal?proteinID=%s&db=%s\">",
displayID, gDatabase);
if (sameWord(spID, displayID) || (strstr(displayID, spID) != NULL))
{
hPrintf("%s</A> %s\n", spID, desc);
}
else
{
hPrintf("%s</A> (aka %s) %s\n", spID, displayID, desc);
}
/* remember the fact that a protein is shown under this PB supported genome */
for (i=0; i<maxPbOrg; i++)
{
if (sameWord(orgSciName, pbOrgSciName[i]))
{
pbOrgPresented[i] = TRUE;
}
}
row = sqlNextRow(sr);
}
hPrintf("</UL>");fflush(stdout);
sqlFreeResult(&sr);
}
sqlDisconnect(&proteinsConn);
row3 = sqlNextRow(srCentral);
}
sqlFreeResult(&srCentral);
hDisconnectCentral(&connCentral);
sqlDisconnect(&conn);
if (protCntInSwissByGene > protCntInSupportedGenomeDb)
{
otherCnt = -1;
if (protCntInSupportedGenomeDb >0)
{
otherCnt = 0;
hPrintf("<FONT SIZE=4><B>Other Organisms:</B></FONT>\n");
hPrintf("<UL>");
}
else
{
hPrintf("<UL>");
}
oldOrg = strdup("");
conn3 = sqlConnect(UNIPROT_DB_NAME);
sqlSafef(query3, sizeof(query3),
"select taxon.id, gene.acc, displayId.val, binomial, description.val "
"from gene, displayId, accToTaxon,taxon, description "
"where gene.val='%s' and gene.acc=displayId.acc and accToTaxon.taxon=taxon.id "
"and accToTaxon.acc=gene.acc and description.acc=gene.acc order by binomial",
queryID);
sr3 = sqlMustGetResult(conn3, query3);
row3 = sqlNextRow(sr3);
/* go through each protein */
while (row3 != NULL)
{
taxonId = row3[0];
protAcc = row3[1];
protDisp = row3[2];
protOrg = row3[3];
protDesc = row3[4];
/* decide if this entry should be skipped */
skipIt = FALSE;
for (i=0; i<maxPbOrg; i++)
{
if (sameWord(pbOrgSciName[i], protOrg) && pbOrgPresented[i])
{
skipIt = TRUE;
}
}
/* print organism name if organism changed */
if (!sameWord(protOrg, oldOrg))
{
if (!sameWord(oldOrg, ""))
{
hPrintf("</UL>\n");
}
if (!skipIt)
{
sqlSafefFrag(cond_str, sizeof(cond_str), "id=%s and nameType='genbank common name'", taxonId);
answer = sqlGetField(PROTEOME_DB_NAME, "taxonNames", "name", cond_str);
hPrintf("<FONT SIZE=3><B>");
hPrintf("<A href=\"http://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?mode=Undef&name=%s&lvl=0&srchmode=1\" TARGET=_blank>%s</A>",
cgiEncode(protOrg), protOrg);
if (answer != NULL)
{
hPrintf(" (%s)", answer);
}
hPrintf(":</B></FONT>\n");
}
hPrintf("<UL>\n");
}
/* print protein entry, if it is not already displayed in the PB supported genome list */
if (!skipIt)
{
otherCnt++;
if (sameWord(protAcc, protDisp))
{
hPrintf("<LI><A HREF=\"../cgi-bin/pbGlobal?proteinID=%s\">", protAcc);
hPrintf("%s</A> %s\n", protAcc, protDesc);
}
else
{
hPrintf("<LI><A HREF=\"../cgi-bin/pbGlobal?proteinID=%s\">", protAcc);
if (strstr(protDisp, protAcc) != NULL)
{
hPrintf("%s</A> %s\n", protAcc, protDesc);
}
else
{
hPrintf("%s</A> (aka %s) %s\n", protAcc, protDisp, protDesc);
}
}
}
oldOrg = strdup(protOrg);
row3 = sqlNextRow(sr3);
}
if (otherCnt == 0) hPrintf("</UL>None");fflush(stdout);
sqlFreeResult(&sr3);
sqlDisconnect(&conn3);
}
}
