void oneTrackInit()
/* Set up global variables using cart settings and initialize libs. */
{
hPrintDisable();
database = cartUsualString(cart, "db", "ce4");
hSetDb(database);
organism = hOrganism(database);
withLeftLabels = FALSE; /* Left labels are not supported. */
withCenterLabels = cartUsualBoolean(cart, "centerLabels", FALSE);
withGuidelines = cartUsualBoolean(cart, "guidelines", FALSE);
revCmplDisp = cartUsualBooleanDb(cart, database, REV_CMPL_DISP, FALSE);
position = cartUsualString(cart, "position", "chrI:4001-5000");
hgParseChromRange(position, &chromName, &winStart, &winEnd);
insideX = 0; /* Left labels are not supported. */
insideWidth = cartUsualInt(cart, "pix", 640);
leftLabelX = 0;
leftLabelWidth = 0; /* Left labels are not supported. */
winBaseCount = winEnd - winStart;
basesPerPixel = ((float)winBaseCount) / ((float)insideWidth);
zoomedToCdsColorLevel = (winBaseCount <= insideWidth*3);
seqBaseCount = hChromSize(database, chromName);
initTl();
zoomedToBaseLevel = (winBaseCount <= insideWidth / tl.mWidth);
zoomedToCodonLevel = (ceil(winBaseCount/3) * tl.mWidth) <= insideWidth;
createHgFindMatchHash();
}
void printOrgAssemblyListAxtInfo(char *dbCgi, char *javascript)
/* Find all the organisms/assemblies that are referenced in axtInfo,
* and print the dropdown list. */
{
struct dbDb *dbList = hGetAxtInfoDbs(dbCgi);
char *assemblyList[128];
char *values[128];
int numAssemblies = 0;
struct dbDb *cur = NULL;
char *assembly = cgiOptionalString(dbCgi);
char orgAssembly[256];
for (cur = dbList; ((cur != NULL) && (numAssemblies < 128)); cur = cur->next)
{
safef(orgAssembly, sizeof(orgAssembly), "%s %s",
cur->organism, cur->description);
assemblyList[numAssemblies] = cloneString(orgAssembly);
values[numAssemblies] = cur->name;
numAssemblies++;
}
#ifdef OLD
// Have to use the "menu" name, not the value, to mark selected:
if (assembly != NULL)
{
char *selOrg = hOrganism(assembly);
char *selFreeze = hFreezeFromDb(assembly);
safef(orgAssembly, sizeof(orgAssembly), "%s %s", selOrg, selFreeze);
assembly = cloneString(orgAssembly);
}
#endif /* OLD */
cgiMakeDropListFull(dbCgi, assemblyList, values, numAssemblies, assembly,
javascript);
}
void addAssemblyToSupportedList(struct dyString* dy, char* assembly)
{
char* organism = hOrganism(assembly);
char* freezeDate = hFreezeDate(assembly);
dyStringPrintf(dy, "%s %s", organism, freezeDate);
freeMem(organism);
freeMem(freezeDate);
}
static void displayRetroDetails(struct sqlConnection *conn, struct mappingInfo *mi)
/* display information from a retroXXXInfo table */
{
struct ucscRetroInfo *pg = mi->pg;
char query[256];
char orthoTable[128];
char orgDb[128];
char *org;
if (mi->suffix != NULL && strlen(mi->suffix) > 0)
safef(orthoTable, sizeof(orthoTable), "%s%sOrtho%s",
mi->tblPre, mi->geneSet, mi->suffix);
else
safef(orthoTable, sizeof(orthoTable), "%s%sOrtho",
mi->tblPre, mi->geneSet);
printf("<TABLE class=\"transMap\">\n");
printf("<H3><A HREF=\"#orthology\">Break in Orthology:</A></H3>\n");
printf("<THEAD>\n");
printf("<TR><TH>Organism<TH>Score</TR>\n");
printf("</THEAD><TBODY>\n");
if (hTableExists(database, orthoTable))
{
struct sqlResult *sr;
char **row;
sqlSafef(query, sizeof(query), "select * from %s where name = '%s' ",
orthoTable, pg->name);
sr = sqlGetResult(conn, query);
while ((row = sqlNextRow(sr)) != NULL)
{
struct ucscRetroOrtho *puro = ucscRetroOrthoLoad(row);
/* get substring after "net" prefix and convert first char to lower
case then get organism name */
safecpy(orgDb, sizeof(orgDb), puro->db+3);
orgDb[0] = tolower(orgDb[0]);
org = hOrganism(orgDb);
printf("<TR><TH>%s (%s) ", org, orgDb);
printf("<TD>%d</TR>\n", puro->overlap);
}
sqlFreeResult(&sr);
}
else
printf("<TR><TH>table %s not found </TR>", orthoTable);
printf("</TBODY></TABLE>\n");
}
static void getItemLabel(struct sqlConnection *conn, char *transMapInfoTbl,
struct sqlConnection *geneConn, char *transMapGeneTbl,
unsigned labelSet,
struct linkedFeatures *lf)
/* get label for a transMap item */
{
struct transMapInfo *info = NULL;
struct transMapGene *gene = NULL;
boolean srcDbExists = FALSE;
if (labelSet & (useOrgCommon|useOrgAbbrv|useOrgDb|useGene))
{
info = transMapInfoQuery(conn, transMapInfoTbl, lf->name);
srcDbExists = sqlDatabaseExists(info->srcDb);
}
if ((labelSet & useGene) && (geneConn != NULL))
{
gene = transMapGeneQuery(geneConn, transMapGeneTbl,
info->srcDb, transMapIdToAcc(info->srcId));
}
struct dyString *label = dyStringNew(64);
if (labelSet & useOrgAbbrv && srcDbExists)
addToLabel(label, orgShortForDb(info->srcDb));
if (labelSet & useOrgCommon && srcDbExists)
addToLabel(label, hOrganism(info->srcDb));
if (labelSet & useOrgDb)
addToLabel(label, info->srcDb);
if (labelSet & useGene)
{
if ((gene != NULL) && (strlen(gene->geneName) > 0))
addToLabel(label, gene->geneName);
else
labelSet |= useAcc; // no gene, so force acc
}
if (labelSet & useAcc)
addToLabel(label, transMapIdToAcc(lf->name));
transMapInfoFree(&info);
transMapGeneFree(&gene);
lf->extra = dyStringCannibalize(&label);
}
boolean doGetBedOrCt(struct sqlConnection *conn, boolean doCt,
boolean doCtFile, boolean redirectToGb)
/* Actually output bed or custom track. Return TRUE unless no results. */
{
char *db = cloneString(database);
char *table = curTable;
struct hTableInfo *hti = getHti(db, table, conn);
struct featureBits *fbList = NULL, *fbPtr;
struct customTrack *ctNew = NULL;
boolean doCtHdr = (cartUsualBoolean(cart, hgtaPrintCustomTrackHeaders, FALSE)
|| doCt || doCtFile);
char *ctWigOutType = cartCgiUsualString(cart, hgtaCtWigOutType, outWigData);
char *fbQual = fbOptionsToQualifier();
char fbTQ[128];
int fields = hTableInfoBedFieldCount(hti);
boolean gotResults = FALSE;
struct region *region, *regionList = getRegions();
boolean isBedGr = isBedGraph(curTable);
boolean isBgWg = isBigWigTable(curTable);
boolean needSubtrackMerge = anySubtrackMerge(database, curTable);
boolean doDataPoints = FALSE;
boolean isWig = isWiggle(database, table);
struct wigAsciiData *wigDataList = NULL;
struct dataVector *dataVectorList = NULL;
boolean doRgb = bedItemRgb(hTrackDbForTrack(db, curTable));
if (!cartUsualBoolean(cart, hgtaDoGreatOutput, FALSE) && !doCt)
{
textOpen();
}
if (cartUsualBoolean(cart, hgtaDoGreatOutput, FALSE))
fputs("#", stdout);
if ((isWig || isBedGr || isBgWg) && sameString(outWigData, ctWigOutType))
doDataPoints = TRUE;
for (region = regionList; region != NULL; region = region->next)
{
struct bed *bedList = NULL, *bed;
struct lm *lm = lmInit(64*1024);
struct dataVector *dv = NULL;
if (isWig && doDataPoints)
{
if (needSubtrackMerge)
{
dv = wiggleDataVector(curTrack, curTable, conn, region);
if (dv != NULL)
slAddHead(&dataVectorList, dv);
}
else
{
int count = 0;
struct wigAsciiData *wigData = NULL;
struct wigAsciiData *asciiData;
struct wigAsciiData *next;
wigData = getWiggleAsData(conn, curTable, region);
for (asciiData = wigData; asciiData; asciiData = next)
{
next = asciiData->next;
if (asciiData->count)
{
slAddHead(&wigDataList, asciiData);
++count;
}
}
slReverse(&wigDataList);
}
}
else if (isBedGr && doDataPoints)
{
dv = bedGraphDataVector(curTable, conn, region);
if (dv != NULL)
slAddHead(&dataVectorList, dv);
}
else if (isBgWg && doDataPoints)
{
dv = bigWigDataVector(curTable, conn, region);
if (dv != NULL)
slAddHead(&dataVectorList, dv);
}
else if (isWig || isBgWg)
{
dv = wiggleDataVector(curTrack, curTable, conn, region);
bedList = dataVectorToBedList(dv);
dataVectorFree(&dv);
}
else if (isBedGr)
{
bedList = getBedGraphAsBed(conn, curTable, region);
}
else
{
bedList = cookedBedList(conn, curTable, region, lm, &fields);
}
/* this is a one-time only initial creation of the custom track
* structure to receive the results. gotResults turns it off after
* the first time.
*/
if (doCtHdr && !gotResults &&
((bedList != NULL) || (wigDataList != NULL) ||
(dataVectorList != NULL)))
{
ctNew = beginCustomTrack(table, fields,
doCt, (isWig || isBedGr || isBgWg), doDataPoints);
}
if (doDataPoints && (wigDataList || dataVectorList))
gotResults = TRUE;
else
{
if ((fbQual == NULL) || (fbQual[0] == 0))
{
for (bed = bedList; bed != NULL; bed = bed->next)
{
if (bed->name != NULL)
{
subChar(bed->name, ' ', '_');
}
if (doCt)
{
struct bed *dupe = cloneBed(bed); /* Out of local memory. */
slAddHead(&ctNew->bedList, dupe);
}
else
{
if (doRgb)
bedTabOutNitemRgb(bed, fields, stdout);
else
bedTabOutN(bed, fields, stdout);
}
gotResults = TRUE;
}
}
else
{
safef(fbTQ, sizeof(fbTQ), "%s:%s", hti->rootName, fbQual);
fbList = fbFromBed(db, fbTQ, hti, bedList, 0, 0, FALSE, FALSE);
if (fields >= 6)
fields = 6;
else if (fields >= 4)
fields = 4;
else
fields = 3;
if (doCt && ctNew)
{
ctNew->fieldCount = fields;
safef(ctNew->tdb->type, strlen(ctNew->tdb->type)+1,
"bed %d", fields);
}
for (fbPtr=fbList; fbPtr != NULL; fbPtr=fbPtr->next)
{
if (fbPtr->name != NULL)
{
char *ptr = strchr(fbPtr->name, ' ');
if (ptr != NULL)
*ptr = 0;
}
if (doCt)
{
struct bed *fbBed = fbToBedOne(fbPtr);
slAddHead(&ctNew->bedList, fbBed );
}
else
{
if (fields >= 6)
hPrintf("%s\t%d\t%d\t%s\t%d\t%c\n",
fbPtr->chrom, fbPtr->start, fbPtr->end, fbPtr->name,
0, fbPtr->strand);
else if (fields >= 4)
hPrintf("%s\t%d\t%d\t%s\n",
fbPtr->chrom, fbPtr->start, fbPtr->end, fbPtr->name);
else
hPrintf("%s\t%d\t%d\n",
fbPtr->chrom, fbPtr->start, fbPtr->end);
}
gotResults = TRUE;
}
featureBitsFreeList(&fbList);
}
}
bedList = NULL;
lmCleanup(&lm);
}
if (!gotResults)
{
hPrintf(NO_RESULTS);
}
else if (doCt)
{
int wigDataSize = 0;
/* Load existing custom tracks and add this new one: */
struct customTrack *ctList = getCustomTracks();
removeNamedCustom(&ctList, ctNew->tdb->table);
if (doDataPoints)
{
if (needSubtrackMerge || isBedGr || isBgWg)
{
slReverse(&dataVectorList);
wigDataSize = dataVectorWriteWigAscii(dataVectorList, ctNew->wigAscii, 0, NULL);
// TODO: see if can make prettier wig output here that
// doesn't necessarily have one value per base
}
else
{
struct wiggleDataStream *wds = NULL;
/* create an otherwise empty wds so we can print out the list */
wds = wiggleDataStreamNew();
wds->ascii = wigDataList;
wigDataSize = wds->asciiOut(wds, db, ctNew->wigAscii, TRUE, FALSE);
#if defined(DEBUG) /* dbg */
/* allow file readability for debug */
chmod(ctNew->wigAscii, 0666);
#endif
wiggleDataStreamFree(&wds);
}
}
else
slReverse(&ctNew->bedList);
slAddHead(&ctList, ctNew);
/* Save the custom tracks out to file (overwrite the old file): */
customTracksSaveCart(db, cart, ctList);
/* Put up redirect-to-browser page. */
if (redirectToGb)
{
char browserUrl[256];
char headerText[512];
int redirDelay = 3;
safef(browserUrl, sizeof(browserUrl),
"%s?%s&db=%s", hgTracksName(), cartSidUrlString(cart), database);
safef(headerText, sizeof(headerText),
"<META HTTP-EQUIV=\"REFRESH\" CONTENT=\"%d;URL=%s\">",
redirDelay, browserUrl);
webStartHeader(cart, database, headerText,
"Table Browser: %s %s: %s", hOrganism(database),
freezeName, "get custom track");
if (doDataPoints)
{
hPrintf("There are %d data points in custom track. ", wigDataSize);
}
else
{
hPrintf("There are %d items in custom track. ",
slCount(ctNew->bedList));
}
hPrintf("You will be automatically redirected to the genome browser in\n"
"%d seconds, or you can \n"
"<A HREF=\"%s\">click here to continue</A>.\n",
redirDelay, browserUrl);
}
}
else if (doDataPoints)
{
if (needSubtrackMerge || isBedGr || isBgWg)
{
slReverse(&dataVectorList);
dataVectorWriteWigAscii(dataVectorList, "stdout", 0, NULL);
}
else
{
/* create an otherwise empty wds so we can print out the list */
struct wiggleDataStream *wds = NULL;
wds = wiggleDataStreamNew();
wds->ascii = wigDataList;
wds->asciiOut(wds, db, "stdout", TRUE, FALSE);
wiggleDataStreamFree(&wds);
}
}
return gotResults;
}
void snpValid()
/* Test snpMap --> dbSnpRs/affy for one assembly. */
{
char *Org;
char *dbSnpTbl = NULL;
struct dbSnpRs *dbSnps = NULL;
struct dbSnpRs *dbSnp = NULL;
struct affy10KDetails *affy10s = NULL;
struct affy10KDetails *affy10 = NULL;
struct affy120KDetails *affy120s = NULL;
struct affy120KDetails *affy120 = NULL;
struct axtScoreScheme *simpleDnaScheme = NULL;
int match = 0; /* good match of minimal acceptable quality */
int mismatch = 0; /* unacceptable match quality */
int missing = 0; /* unable to find rsId in dbSnpRs/affy */
int goodrc = 0; /* matches after reverse-complement */
int assemblyDash = 0; /* assembly context is just a single dash - (complex cases) */
int gapNib = 0; /* nib returns n's, we are in the gap */
int totalMatch = 0;
int totalMismatch = 0;
int totalMissing = 0;
int totalGoodrc = 0;
int totalAssemblyDash = 0;
int totalGapNib = 0;
boolean affy = FALSE;
int mode = 3;
void *next = NULL;
char *id = NULL;
char *seq = NULL;
char affy120id[12];
int matchScore = 100;
int misMatchScore = 100;
int gapOpenPenalty = 400;
int gapExtendPenalty = 50;
int noDna = 0;
int snpMapRows = 0;
/* controls whether affy120k, affy10k, or dbSnpRs is used
currently affys are human only
*/
if (!hDbIsActive(db))
{
printf("Currently no support for db %s\n", db);
return;
}
hSetDb(db);
Org = hOrganism(db);
if (sameWord(Org,"Human"))
affy = TRUE;
if (sameWord(Org,"Human"))
dbSnpTbl = "dbSnpRsHg";
else if (sameWord(Org,"Mouse"))
dbSnpTbl = "dbSnpRsMm";
else if (sameWord(Org,"Rat"))
dbSnpTbl = "dbSnpRsRn";
else
{
printf("Currently no support for Org %s\n", Org);
return;
}
simpleDnaScheme = axtScoreSchemeSimpleDna(matchScore, misMatchScore, gapOpenPenalty, gapExtendPenalty);
uglyf("dbSnp Table=%s \n",dbSnpTbl);
uglyf("Affy=%s \n", affy ? "TRUE" : "FALSE" );
dbSnps = readDbSnps(dbSnpTbl);
printf("read hgFixed.%s \n",dbSnpTbl);
if (affy)
{
affy10s = readAffy10();
printf("read hgFixed.affy10KDetails \n");
affy120s = readAffy120();
printf("read hgFixed.affy120KDetails \n");
}
int bogus = 0;
// debug
if (0)
{
printf("rsId assembly-sequence \n");
printf("---------------------------------------------- \n");
for (dbSnp = dbSnps; dbSnp != NULL; dbSnp = dbSnp->next)
{
printf("%s %s \n",
dbSnp->rsId,
dbSnp->assembly
);
// debug: cut it short for testing only
if (++bogus > 1)
break;
}
printf("\n");
printf("\n");
}
bogus=0;
struct slName *cns = hAllChromNames();
struct slName *cn=NULL;
if (!cns)
{
printf("testDb: hAllChromNames returned empty list \n");
return;
}
if (affy)
{
mode=1; /* start on affy120 with numbers in snpMap.rsId */
}
else
{
mode=2; /* start on dbSnps with "rs*" in snpMap.rsId */
}
for (cn = cns; cn != NULL; cn = cn->next)
{
struct dnaSeq *chromSeq = NULL;
struct snpMap *snps = NULL;
struct snpMap *snp = NULL;
if (chr != NULL)
if (!sameWord(chr,cn->name))
continue;
//uglyf("testDb: beginning chrom %s \n",cn->name);
chromSeq = hLoadChrom(cn->name);
printf("testDb: chrom %s : size (%u) \n",cn->name,chromSeq->size);
snps = readSnps(cn->name);
printf("read %s.snpMap where chrom=%s \n",db,cn->name);
dbSnp = dbSnps;
affy10 = affy10s;
affy120 = affy120s;
printf("=========================================================\n");
for (snp = snps; snp != NULL; snp = snp->next)
{
int cmp = -1;
char *nibDna=NULL;
char *nibDnaRc=NULL;
++snpMapRows;
/*
printf("%s %s %u %u %s\n",
snp->name,
snp->chrom,
snp->chromStart,
snp->chromEnd,
nibDna
);
*/
while (cmp < 0)
{
while (cmp < 0)
{
switch (mode)
{
case 1:
next = affy120; break;
case 2:
next = dbSnp; break;
case 3:
next = affy10; break;
}
if (next == NULL)
{
switch (mode)
{
case 1:
++mode; break;
case 2:
++mode; break;
case 3:
cmp = 1; break;
}
}
else
{
break;
}
}
if (cmp < 0)
{
switch (mode)
{
case 1:
safef(affy120id, sizeof(affy120id), "%d", affy120->affyId); /* have int type but want string */
id = affy120id;
break;
case 2:
id = dbSnp->rsId; break;
case 3:
id = affy10->affyId; break;
}
cmp=mystrcmp(id, snp->name);
}
if (cmp < 0)
{
switch (mode)
{
case 1:
affy120 = affy120->next; break;
case 2:
dbSnp = dbSnp->next; break;
case 3:
affy10 = affy10->next; break;
}
}
}
if (cmp==0)
{
int strand=1;
char *rc = NULL;
int m = 0;
int lf = 0; /* size of left flank context (lower case dna) */
int rf = 0; /* size of right flank context (lower case dna) */
int ls = 0; /* total size of assembly dna context plus actual region in dbSnpRs/affy */
char *origSeq = NULL; /* use to display the original dnSnpRs.assembly seq */
switch (mode)
{
case 1:
seq = affy120->sequenceA; break;
case 2:
seq = dbSnp->assembly; break;
case 3:
seq = affy10->sequenceA; break;
}
if (sameString(seq,"-"))
{
++assemblyDash;
if (Verbose)
printf("(no assembly context) rsId=%s chrom=%s %u %u \n assembly=%s \n\n",
id,
snp->chrom,
snp->chromStart,
snp->chromEnd,
seq
);
continue;
}
origSeq = seq;
lf = leftFlank(origSeq);
rf = rightFlank(origSeq);
seq = cloneString(origSeq);
stripDashes(seq); /* remove dashes indicating insert to simplify and correct processing of nib data */
ls = strlen(seq); /* used to be: lengthOneDash(seq); */
//debug
//uglyf("about to call checkandFetchNib origSeq=%s lf=%d, rf=%d ls=%d \n", origSeq, lf, rf, ls);
nibDna = checkAndFetchNib(chromSeq, snp, lf, ls);
if (nibDna==NULL)
{
++noDna;
printf("no dna for %s %s %u %u \n",
snp->name,
snp->chrom,
snp->chromStart,
snp->chromEnd
);
continue;
}
//debug
//uglyf("got past checkandFetchNib call: \n nibDna=%s \n",nibDna);
if (allNs(nibDna))
{
++gapNib;
++mismatch;
if (Verbose)
printf("(nib gap) rsId=%s chrom=%s %u %u \n assembly=%s \n snpMap=%s \n\n",
id,
snp->chrom,
snp->chromStart,
snp->chromEnd,
seq,
nibDna
);
continue;
}
m = misses(seq,nibDna);
if (m > 1)
{
//debug
//uglyf("rc: about to call checkandFetchNib \n");
rc = checkAndFetchNib(chromSeq, snp, rf, ls);
if (rc==NULL)
{
++noDna;
printf("no dna for %s %s %u %u \n",
snp->name,
snp->chrom,
snp->chromStart,
snp->chromEnd
);
continue;
}
//debug
//uglyf("rc: got past checkandFetchNib call: \n rc Dna=%s \n",rc);
reverseComplement(rc,strlen(rc));
int n = misses(seq, rc);
if (n < m)
{
strand=-1;
m = n;
}
}
if (m <= 1)
{
++match;
if (strand < 1)
++goodrc;
}
else
{
struct dnaSeq query, target;
struct axt *axtAln = NULL;
int bestScore = 0;
ZeroVar(&query);
query.dna = seq;
query.size = strlen(query.dna);
ZeroVar(&target);
target.dna = nibDna;
target.size = strlen(target.dna);
axtAln = axtAffine(&query, &target, simpleDnaScheme);
strand = 1;
if (axtAln)
{
bestScore = axtAln->score / ls;
}
axtFree(&axtAln);
if (bestScore < threshold)
{
ZeroVar(&target);
target.dna = rc;
target.size = strlen(target.dna);
axtAln = axtAffine(&query, &target, simpleDnaScheme);
if ((axtAln) && (bestScore < (axtAln->score / ls)))
{
strand = -1;
bestScore = axtAln->score / ls;
}
axtFree(&axtAln);
}
if (bestScore >= threshold)
{
++match;
if (strand < 1)
++goodrc;
}
else
{
++mismatch;
}
if ((bestScore < threshold) || Verbose)
{
printf(
"score=%d misses=%u strand=%d rsId=%s chrom=%s %u %u lf=%d ls=%d \n"
" assembly=%s \n"
" snpMap=%s \n"
"rc snpMap=%s \n"
"\n",
bestScore,
m,
strand,
id,
snp->chrom,
snp->chromStart,
snp->chromEnd,
lf,
ls,
seq,
nibDna,
rc
);
}
}
freez(&rc);
freez(&seq);
}
else
{
char snpLkup[10] = "";
/* this id is missing from dbSnpRs/affy! */
++missing;
switch (mode)
{
case 1:
safef(snpLkup,sizeof(snpLkup),"%s","affy120"); break;
case 2:
safef(snpLkup,sizeof(snpLkup),"%s",dbSnpTbl); break;
case 3:
safef(snpLkup,sizeof(snpLkup),"%s","affy10"); break;
}
if (Verbose)
printf("snpMap.name=%s is missing from %s (now at %s) \n\n",snp->name,snpLkup,id);
}
freez(&nibDna);
// debug: cut it short for testing only
//break;
}
snpMapFreeList(&snps);
dnaSeqFree(&chromSeq);
printf("\n\n\n Total matches for chrom %s:\n ",cn->name);
printf(" matches: %u \n ",match);
printf(" mismatches: %u \n",mismatch);
printf("missing from dbSnpRs: %u \n",missing);
printf(" rev compl matches: %u \n",goodrc);
printf(" assembly = -: %u \n",assemblyDash);
printf(" nib in gap : %u \n",gapNib);
printf("\n\n=========================================\n");
totalMatch += match;
totalMismatch += mismatch;
totalMissing += missing;
totalGoodrc += goodrc;
totalAssemblyDash += assemblyDash;
totalGapNib += gapNib;
match = 0;
mismatch = 0;
missing = 0;
goodrc = 0;
assemblyDash = 0;
gapNib = 0;
// debug: cut it to just one or two chrom for testing
//if (++bogus > 1)
// break;
printf("\n");
printf("\n");
}
slFreeList(&cns);
dbSnpRsFreeList(&dbSnps);
if (affy)
{
affy10KDetailsFreeList(&affy10s);
affy120KDetailsFreeList(&affy120s);
}
axtScoreSchemeFree(&simpleDnaScheme);
printf("\n\n\n Grand Totals: \n ");
printf(" matches: %u \n ",totalMatch);
printf(" mismatches: %u \n",totalMismatch);
printf("missing from dbSnpRs: %u \n",totalMissing);
printf(" rev compl matches: %u \n",totalGoodrc);
printf(" assembly = -: %u \n",totalAssemblyDash);
printf(" nib in gap : %u \n",totalGapNib);
printf("\n Total rows in snpMap: %u \n ",snpMapRows);
printf("\n # no dna found for : %u \n ",noDna);
printf("\n\n=========================================\n");
}
int main(int argc, char *argv[])
/* Process command line. */
{
optionInit(&argc, argv, options);
if (optionExists("help"))
{
printHelp();
}
if (argc != 3)
{
usage();
}
outMdb = optionVal("outMdb", outMdb);
onlyCompTdb = optionExists("onlyCompTdb");
release = optionVal("release", release);
releaseNum = optionInt("releaseNum", releaseNum);
char *database = argv[1];
char *composite = argv[2];
char defaultMetaDb[1024];
char defaultDownloadDir[1024];
char tempDownloadDir[1024];
char *src = getSrcDir();
char *org = cloneString(hOrganism(database));
org[0] = tolower(org[0]);
/* If user doesn't provide a metaDB, assume the path using the database and composite */
safef(defaultMetaDb, sizeof(defaultMetaDb), "%s/hg/makeDb/trackDb/%s/%s/metaDb/%s/%s.ra", src, org, database, release, composite);
/* If user doesn't provide a downloadDir, assume the path using the database and composite */
safef(defaultDownloadDir, sizeof(defaultDownloadDir), "/usr/local/apache/htdocs-hgdownload/goldenPath/%s/encodeDCC/%s", database, composite);
safef(tempDownloadDir, sizeof(tempDownloadDir), "/usr/local/apache/htdocs-hgdownload/goldenPath/%s/encodeDCC/%s", database, composite);
if (releaseNum)
safef(defaultDownloadDir, sizeof(defaultDownloadDir), "%s/release%d", tempDownloadDir, releaseNum);
/* If user doesn't provide a trackDB, assume the path using the database and composite */
char defaultTrackDb[1024];
/* Load encode composite-includer trackDb.wgEncode.ra */
char trackDbIncluder[1024];
safef(trackDbIncluder, sizeof(trackDbIncluder), "%s/hg/makeDb/trackDb/%s/%s/%s", src, org, database, "trackDb.wgEncode.ra");
struct raFile *includerFile = raFileRead(trackDbIncluder);
/* Find the correct trackDb.ra for the composite */
int numTagsFound = -1;
char *compositeName = findCompositeRa(includerFile, composite, release, &numTagsFound);
if (!compositeName)
errAbort("unable to find composite .ra for the track in trackDb.wgEncode.ra\n");
// if numTagsFound == 1 then a composite .ra with a single alpha tag exists already,
// so no further work required on trackDb.wgEncode.ra
safef(defaultTrackDb, sizeof(defaultTrackDb), "%s/hg/makeDb/trackDb/%s/%s/%s", src, org, database, compositeName);
verbose(1,"database: %s\ncomposite: %s\nrelease %s\ndefault trackDb: %s\ndefault metaDb: %s\ndefault downloadDir: %s\n",
database, composite, release, defaultTrackDb, defaultMetaDb, defaultDownloadDir);
char *metaDb = optionVal("metaDb",defaultMetaDb);
replaceTildeWithHome(&metaDb);
if (!fileExists(metaDb))
errAbort("metaDb %s does not exist.", metaDb);
char *trackDb = optionVal("trackDb",defaultTrackDb);
replaceTildeWithHome(&trackDb);
if (!fileExists(trackDb))
errAbort("trackDb %s does not exist.", trackDb);
char *downloadDir = optionVal("downloadDir",defaultDownloadDir);
replaceTildeWithHome(&downloadDir);
if (!fileExists(downloadDir))
errAbort("downloadDir %s does not exist.", downloadDir);
printf("metaDb = %s\n trackDb = %s\n downloadDir = %s\n",metaDb,trackDb,downloadDir);
metaCheck(database, composite, metaDb, trackDb, downloadDir);
return 0;
}
void mafPrettyOut(FILE *f, struct mafAli *maf, int lineSize,
boolean onlyDiff, int blockNo)
{
int ii, ch;
int srcChars = 0;
struct mafComp *mc;
int lineStart, lineEnd;
char *summaryLine = needMem(lineSize+1);
char *referenceText;
int startChars, sizeChars, srcSizeChars;
boolean haveInserts = FALSE;
struct mafComp *masterMc = maf->components;
startChars = sizeChars = srcSizeChars = 0;
for (mc = maf->components; mc != NULL; mc = mc->next)
{
/* Figure out length of source (species) field. */
/*if (mc->size != 0)*/
{
char dbOnly[128];
int len;
char *org;
memset(dbOnly, 0, sizeof(dbOnly));
safef(dbOnly, sizeof(dbOnly), "%s", mc->src);
chopPrefix(dbOnly);
if ((org = hOrganism(dbOnly)) == NULL)
len = strlen(dbOnly);
else
len = strlen(org);
if (srcChars < len)
srcChars = len;
len = digitsBaseTen(mc->start);
if (startChars < len)
startChars = len;
len = digitsBaseTen(mc->size);
if (sizeChars < len)
sizeChars = len;
len = digitsBaseTen(mc->srcSize);
if (srcSizeChars < len)
srcSizeChars = len;
if (mc->text && (mc->rightStatus == MAF_INSERT_STATUS) && (masterMc->start + masterMc->size < winEnd))
haveInserts = TRUE;
#ifdef REVERSESTRAND
/* complement bases if hgTracks is on reverse strand */
if (mc->size && cartCgiUsualBoolean(cart, COMPLEMENT_BASES_VAR, FALSE))
complement(mc->text, maf->textSize);
#endif
}
}
/* first sequence in the alignment */
referenceText = maf->components->text;
for (lineStart = 0; lineStart < maf->textSize; lineStart = lineEnd)
{
int size;
lineEnd = lineStart + lineSize;
if (lineEnd >= maf->textSize)
lineEnd = maf->textSize;
size = lineEnd - lineStart;
initSummaryLine(summaryLine, size, '*');
for (mc = maf->components; mc != NULL; mc = mc->next)
{
char dbOnly[128], *chrom;
int s = mc->start;
int e = s + mc->size;
char *org;
char *revComp = "";
char strand = mc->strand;
struct dyString *dy = newDyString(512);
#ifdef REVERSESTRAND
if (cartCgiUsualBoolean(cart, COMPLEMENT_BASES_VAR, FALSE))
strand = (strand == '+') ? '-' : '+';
#endif
if (strand == '-') revComp = "&hgSeq.revComp=on";
dyStringClear(dy);
safef(dbOnly, sizeof(dbOnly), "%s", mc->src);
chrom = chopPrefix(dbOnly);
if ((org = hOrganism(dbOnly)) == NULL)
org = dbOnly;
if (mc->strand == '-')
reverseIntRange(&s, &e, mc->srcSize);
if (mc->text != NULL)
{
if (lineStart == 0)
{
if (hDbIsActive(dbOnly))
{
dyStringPrintf(dy, "%s Browser %s:%d-%d %c %*dbps",hOrganism(dbOnly),chrom, s+1, e, mc->strand,sizeChars, mc->size);
linkToOtherBrowserTitle(dbOnly, chrom, s, e, dy->string);
dyStringClear(dy);
fprintf(f, "B</A> ");
}
else
fprintf(f, " ");
if (hDbExists(dbOnly))
{
dyStringPrintf(dy, "Get %s DNA %s:%d-%d %c %*dbps",hOrganism(dbOnly),chrom, s+1, e, mc->strand,sizeChars, mc->size);
printf("<A TITLE=\"%s\" TARGET=\"_blank\" HREF=\"%s?o=%d&g=getDna&i=%s&c=%s&l=%d&r=%d&db=%s%s\">D</A> ", dy->string,hgcName(),
s, cgiEncode(chrom),
chrom, s, e, dbOnly, revComp);
}
else
fprintf(f, " ");
}
else
{
fprintf(f, " ");
}
dyStringClear(dy);
dyStringPrintf(dy, "%s:%d-%d %c %*dbps",chrom, s+1, e, mc->strand,sizeChars, mc->size);
fprintf(f, "<A TITLE=\"%s\"> %*s </A> ", dy->string, srcChars, org);
updateSummaryLine(summaryLine, referenceText + lineStart, mc->text + lineStart, size);
blueCapWrite(f, mc->text + lineStart, size,
(onlyDiff && mc != maf->components) ? referenceText + lineStart : NULL);
fprintf(f, "\n");
}
else
{
if (((mc->leftStatus == MAF_CONTIG_STATUS) && (mc->rightStatus == MAF_CONTIG_STATUS) )
|| ((mc->leftStatus == MAF_TANDEM_STATUS) && (mc->rightStatus == MAF_TANDEM_STATUS) )
|| ((mc->leftStatus == MAF_INSERT_STATUS) && (mc->rightStatus == MAF_INSERT_STATUS) )
|| ((mc->leftStatus == MAF_MISSING_STATUS) && (mc->rightStatus == MAF_MISSING_STATUS) ))
{
if (lineStart == 0)
{
int s = mc->start;
int e = s + mc->rightLen;
struct dyString *dy = newDyString(512);
if (mc->strand == '-')
reverseIntRange(&s, &e, mc->srcSize);
if ( hDbIsActive(dbOnly))
{
dyStringPrintf(dy, "%s Browser %s:%d-%d %c %d bps Unaligned",hOrganism(dbOnly),chrom, s+1, e, mc->strand, e-s);
linkToOtherBrowserTitle(dbOnly, chrom, s, e, dy->string);
fprintf(f,"B</A> ");
dyStringClear(dy);
}
else
fprintf(f," ");
if (hDbExists(dbOnly))
{
dyStringPrintf(dy, "Get %s DNA %s:%d-%d %c %d bps Unaligned",hOrganism(dbOnly),chrom, s+1, e, mc->strand, e-s);
printf("<A TITLE=\"%s\" TARGET=\"_blank\" HREF=\"%s?o=%d&g=getDna&i=%s&c=%s&l=%d&r=%d&db=%s%s\">D</A> ", dy->string, hgcName(),
s, cgiEncode(chrom),
chrom, s, e, dbOnly,revComp);
}
else
fprintf(f, " ");
}
else
fprintf(f, " ");
initSummaryLine(summaryLine, size, ' ');
dyStringClear(dy);
dyStringPrintf(dy, "%s:%d-%d %c %*dbps",chrom, s+1, e, mc->strand,sizeChars, mc->size);
fprintf(f, "<A TITLE=\"%s\">%*s</A> ", dy->string, srcChars, org);
ch = '-';
switch(mc->rightStatus)
{
case MAF_INSERT_STATUS:
ch = '=';
break;
case MAF_MISSING_STATUS:
ch = 'N';
break;
case MAF_TANDEM_STATUS:
case MAF_CONTIG_STATUS:
ch = '-';
break;
}
for(ii=lineStart; ii < lineEnd ; ii++)
fputc(ch,f);
fprintf(f,"\n");
}
}
}
#ifdef ADDMATCHLINE
if (lineStart == 0)
fprintf(f, " %-*s %s\n", srcChars, "", summaryLine);
else
fprintf(f, "%-*s %s\n", srcChars, "", summaryLine);
#else
fprintf(f, "\n");
#endif
}
if (haveInserts)
{
fprintf(f, "<B>Inserts between block %d and %d in window</B>\n",blockNo, blockNo+1);
for (mc = maf->components; mc != NULL; mc = mc->next)
{
char dbOnly[128], *chrom;
int s = mc->start + mc->size;
int e = s + mc->rightLen;
char *org;
if (mc->text == NULL)
continue;
if (mc->strand == '-')
reverseIntRange(&s, &e, mc->srcSize);
safef(dbOnly, sizeof(dbOnly), "%s", mc->src);
chrom = chopPrefix(dbOnly);
if ((org = hOrganism(dbOnly)) == NULL)
org = dbOnly;
if (mc->rightStatus == MAF_INSERT_STATUS)
{
char *revComp = "";
if (hDbIsActive(dbOnly))
{
char strand = mc->strand;
#ifdef REVERSESTRAND
if (cartCgiUsualBoolean(cart, COMPLEMENT_BASES_VAR, FALSE))
strand = (strand == '+') ? '-' : '+';
#endif
if (strand == '-') revComp = "&hgSeq.revComp=on";
linkToOtherBrowser(dbOnly, chrom, s, e);
fprintf(f,"B");
fprintf(f, "</A>");
fprintf(f, " ");
}
else
fprintf(f, " ");
if (hDbExists(dbOnly))
{
printf("<A TARGET=\"_blank\" HREF=\"%s?o=%d&g=getDna&i=%s&c=%s&l=%d&r=%d"
"&db=%s%s\">D</A> ", hgcName(), s, cgiEncode(chrom), chrom, s,
e, dbOnly,revComp);
}
else
fprintf(f, " ");
fprintf(f, "%*s %dbp\n", srcChars, org,mc->rightLen);
}
}
fprintf(f, "\n");
}
freeMem(summaryLine);
}
static void mafOrAxtClick2(struct sqlConnection *conn, struct sqlConnection *conn2, struct trackDb *tdb, char *axtOtherDb, char *fileName)
/* Display details for MAF or AXT tracks. */
{
hgBotDelay();
if (winEnd - winStart > 30000)
{
printf("Zoom so that window is 30,000 bases or less to see alignments and conservation statistics\n");
}
else
{
struct mafAli *mafList = NULL, *maf, *subList = NULL;
int aliIx = 0, realCount = 0;
char dbChrom[64];
char option[128];
char *capTrack;
struct consWiggle *consWig, *consWiggles;
struct hash *speciesOffHash = NULL;
char *speciesOrder = NULL;
char *speciesTarget = trackDbSetting(tdb, SPECIES_TARGET_VAR);
char buffer[1024];
int useTarg = FALSE;
int useIrowChains = FALSE;
safef(option, sizeof(option), "%s.%s", tdb->track, MAF_CHAIN_VAR);
if (cartCgiUsualBoolean(cart, option, FALSE) &&
trackDbSetting(tdb, "irows") != NULL)
useIrowChains = TRUE;
safef(buffer, sizeof(buffer), "%s.vis",tdb->track);
if (useIrowChains)
{
if (!cartVarExists(cart, buffer) && (speciesTarget != NULL))
useTarg = TRUE;
else
{
char *val;
val = cartUsualString(cart, buffer, "useCheck");
useTarg = sameString("useTarg",val);
}
}
if (sameString(tdb->type, "bigMaf"))
{
char *fileName = trackDbSetting(tdb, "bigDataUrl");
struct bbiFile *bbi = bigBedFileOpen(fileName);
mafList = bigMafLoadInRegion(bbi, seqName, winStart, winEnd);
}
else
mafList = mafOrAxtLoadInRegion2(conn,conn2, tdb, seqName, winStart, winEnd,
axtOtherDb, fileName);
safef(dbChrom, sizeof(dbChrom), "%s.%s", hubConnectSkipHubPrefix(database), seqName);
safef(option, sizeof(option), "%s.speciesOrder", tdb->track);
speciesOrder = cartUsualString(cart, option, NULL);
if (speciesOrder == NULL)
speciesOrder = trackDbSetting(tdb, "speciesOrder");
for (maf = mafList; maf != NULL; maf = maf->next)
{
int mcCount = 0;
struct mafComp *mc;
struct mafAli *subset;
struct mafComp *nextMc;
/* remove empty components and configured off components
* from MAF, and ignore
* the entire MAF if all components are empty
* (solely for gap annotation) */
if (!useTarg)
{
for (mc = maf->components->next; mc != NULL; mc = nextMc)
{
char buf[64];
char *organism;
mafSrcDb(mc->src, buf, sizeof buf);
organism = hOrganism(buf);
if (!organism)
organism = buf;
nextMc = mc->next;
safef(option, sizeof(option), "%s.%s", tdb->track, buf);
if (!cartUsualBoolean(cart, option, TRUE))
{
if (speciesOffHash == NULL)
speciesOffHash = newHash(4);
hashStoreName(speciesOffHash, organism);
}
if (!cartUsualBoolean(cart, option, TRUE))
slRemoveEl(&maf->components, mc);
else
mcCount++;
}
}
if (mcCount == 0)
continue;
if (speciesOrder)
{
int speciesCt;
char *species[2048];
struct mafComp **newOrder, *mcThis;
int i;
mcCount = 0;
speciesCt = chopLine(cloneString(speciesOrder), species);
newOrder = needMem((speciesCt + 1) * sizeof (struct mafComp *));
newOrder[mcCount++] = maf->components;
for (i = 0; i < speciesCt; i++)
{
if ((mcThis = mafMayFindCompSpecies(maf, species[i], '.')) == NULL)
continue;
newOrder[mcCount++] = mcThis;
}
maf->components = NULL;
for (i = 0; i < mcCount; i++)
{
newOrder[i]->next = 0;
slAddHead(&maf->components, newOrder[i]);
}
slReverse(&maf->components);
}
subset = mafSubsetE(maf, dbChrom, winStart, winEnd, TRUE);
if (subset != NULL)
{
/* Reformat MAF if needed so that sequence from current
* database is the first component and on the
* plus strand. */
mafMoveComponentToTop(subset, dbChrom);
if (subset->components->strand == '-')
mafFlipStrand(subset);
subset->score = mafScoreMultiz(subset);
slAddHead(&subList, subset);
++realCount;
}
}
slReverse(&subList);
mafAliFreeList(&mafList);
if (subList != NULL)
{
char *showVarName = "hgc.showMultiBase";
char *showVarVal = cartUsualString(cart, showVarName, "all");
boolean onlyDiff = sameWord(showVarVal, "diff");
#ifdef ADDEXONCAPITAL
char *codeVarName = "hgc.multiCapCoding";
char *codeVarVal = cartUsualString(cart, codeVarName, "coding");
boolean onlyCds = sameWord(codeVarVal, "coding");
#endif
/* add links for conservation score statistics */
consWiggles = wigMafWiggles(database, tdb);
int wigCount = slCount(consWiggles);
if (wigCount == 1)
{
conservationStatsLink(tdb, "Conservation score statistics", consWiggles->table);
}
else if (wigCount > 1)
{
/* multiple wiggles. List all that have been turned on with
* checkboxes */
/* Scan for cart variables -- do any exist, are any turned on ? */
boolean wigSet = FALSE;
boolean wigOn = FALSE;
for (consWig = consWiggles; consWig != NULL;
consWig = consWig->next)
{
char *wigVarSuffix = NULL;
(void)wigMafWiggleVar(tdb->track, consWig, &wigVarSuffix);
if (cartVarExistsAnyLevel(cart, tdb, FALSE, wigVarSuffix))
{
wigSet = TRUE;
if (cartBooleanClosestToHome(cart, tdb, FALSE, wigVarSuffix))
wigOn = TRUE;
}
}
/* If there are no cart vars, turn on the first (default) wig */
if (!wigSet)
{
char *prefix = tdb->track; // use when setting things to the cart
if (tdbIsContainerChild(tdb))
prefix = tdbGetContainer(tdb)->track;
cartSetBoolean(cart, wigMafWiggleVar(prefix, consWiggles, NULL), TRUE);
wigOn = TRUE;
}
if (wigOn)
{
boolean first = TRUE;
for (consWig = consWiggles; consWig != NULL;
consWig = consWig->next)
{
if (first)
{
printf("Conservation score statistics:");
first = FALSE;
}
char *wigVarSuffix = NULL;
(void)wigMafWiggleVar(tdb->track, consWig, &wigVarSuffix);
if (cartUsualBooleanClosestToHome(cart, tdb, FALSE, wigVarSuffix,FALSE))
{
printf(" ");
subChar(consWig->uiLabel, '_', ' ');
conservationStatsLink(tdb,
consWig->uiLabel, consWig->table);
}
}
}
}
puts("</P>\n");
/* no alignment to display when in visibilities where only wiggle is shown */
char *vis = cartOptionalString(cart, tdb->track);
if (vis)
{
enum trackVisibility tv = hTvFromStringNoAbort(vis);
if (tv == tvSquish || tv == tvDense)
return;
}
#ifdef ADDEXONCAPITAL
puts("<FORM ACTION=\"../cgi-bin/hgc\" NAME=\"gpForm\" METHOD=\"GET\">");
cartSaveSession(cart);
cgiContinueHiddenVar("g");
cgiContinueHiddenVar("c");
cgiContinueHiddenVar("i");
printf("Capitalize ");
cgiMakeDropListFull(codeVarName, codeAll, codeAll,
ArraySize(codeAll), codeVarVal, autoSubmit);
printf("exons based on ");
capTrack = genePredDropDown(cart, trackHash,
"gpForm", "hgc.multiCapTrack");
#endif
printf("show ");
cgiMakeDropListFull(showVarName, showAll, showAll,
ArraySize(showAll), showVarVal, autoSubmit);
printf("bases");
printf("<BR>\n");
printf("</FORM>\n");
#ifdef REVERSESTRAND
/* notify if bases are complemented (hgTracks is on reverse strand) */
if (cartCgiUsualBoolean(cart, COMPLEMENT_BASES_VAR, FALSE))
puts("<EM>Alignment displayed on reverse strand</EM><BR>");
#endif
puts("Place cursor over species for alignment detail. Click on 'B' to link to browser ");
puts("for aligned species, click on 'D' to get DNA for aligned species.<BR>");
printf("<TT><PRE>");
/* notify if species removed from alignment */
if (speciesOffHash)
{
char *species;
struct hashCookie hc = hashFirst(speciesOffHash);
puts("<B>Components not displayed:</B> ");
while ((species = hashNextName(&hc)) != NULL)
printf("%s ", species);
puts("<BR>");
}
for (maf = subList; maf != NULL; maf = maf->next)
{
mafLowerCase(maf);
#ifdef ADDEXONCAPITAL
if (capTrack != NULL)
capMafOnTrack(maf, capTrack, onlyCds);
#endif
printf("<B>Alignment block %d of %d in window, %d - %d, %d bps </B>\n",
++aliIx,realCount,maf->components->start + 1,
maf->components->start + maf->components->size, maf->components->size);
mafPrettyOut(stdout, maf, 70,onlyDiff, aliIx);
}
mafAliFreeList(&subList);
}
else
{
printf("No multiple alignment in browser window");
}
printf("</PRE></TT>");
}
}
