static void altSplicePrint(struct section *section,
struct sqlConnection *conn, char *geneId)
/* Print out altSplicing info. */
{
char *altId = section->items;
char query[256];
struct sqlResult *sr;
char **row;
struct altGraphX *ag;
char table[64];
boolean hasBin;
hFindSplitTable(sqlGetDatabase(conn), curGeneChrom, "altGraphX", table, &hasBin);
sqlSafef(query, sizeof(query), "select * from %s where name='%s'", table, altId);
sr = sqlGetResult(conn, query);
if ((row = sqlNextRow(sr)) != NULL)
{
ag = altGraphXLoad(row+hasBin);
hPrintf("<TABLE><TR><TD BGCOLOR='#888888'>\n");
altGraphXMakeImage(ag);
hPrintf("</TD></TR></TABLE><BR>");
}
sqlFreeResult(&sr);
hPrintf("This graph shows alternative splicing observed in mRNAs and "
"ESTs that is either conserved in mouse, present in full length "
"mRNAs, or observed at least three times in ESTs.");
}
static struct slName *getExamples(char *db, struct sqlConnection *conn,
char *table, char *field, int count)
/* Return a list of several example values of table.field. */
{
boolean isTabix = FALSE;
if (isBamTable(table))
{
assert(sameString(field, "qName"));
return randomBamIds(table, conn, count);
}
else if (isBigBed(db, table, curTrack, ctLookupName))
{
assert(sameString(field, "name"));
return randomBigBedIds(table, conn, count);
}
else if (isVcfTable(table, &isTabix))
{
assert(sameString(field, "id"));
return randomVcfIds(table, conn, count, isTabix);
}
else
{
char fullTable[HDB_MAX_TABLE_STRING];
char *c = strchr(table, '.');
if (c || ! hFindSplitTable(database, NULL, table, fullTable, NULL))
safecpy(fullTable, sizeof(fullTable), table);
return sqlRandomSampleConn(conn, fullTable, field, count);
}
}
void orTable(char *database, Bits *acc, char *track, char *chrom,
int chromSize, struct sqlConnection *conn)
/* Or in table if it exists. Else do nothing. */
{
char t[512], *s;
char table[HDB_MAX_TABLE_STRING];
isolateTrackPartOfSpec(track, t);
s = strrchr(t, '.');
if (s != NULL)
{
orFile(acc, track, chrom, chromSize);
}
else
{
boolean hasBin;
int minFeatureSize = optionInt("minFeatureSize", 0);
boolean isSplit = hFindSplitTable(database, chrom, t, table, &hasBin);
boolean isFound = hTableExists(database, table);
verbose(3,"orTable: db: %s isFound: %s isSplit: %s %s %s %s\n", database,
isFound ? "TRUE" : "FALSE",
isSplit ? "TRUE" : "FALSE", chrom, t, table );
if (isFound)
fbOrTableBitsQueryMinSize(database, acc, track, chrom, chromSize, conn, where,
TRUE, TRUE, minFeatureSize);
}
}
static struct genePred *getCurGenePred(struct sqlConnection *conn)
/* Return current gene in genePred. */
{
char *track = genomeSetting("knownGene");
char table[HDB_MAX_TABLE_STRING];
boolean hasBin;
char query[256];
struct sqlResult *sr;
char **row;
struct genePred *gp = NULL;
if (!hFindSplitTable(sqlGetDatabase(conn), curGeneChrom, track, table, sizeof table, &hasBin))
errAbort("track %s not found", track);
bool hasAttrId = sqlColumnExists(conn, table, "alignId");
sqlSafef(query, sizeof(query),
"select * from %s where name = '%s' "
"and chrom = '%s' and txStart=%d and txEnd=%d"
, table, curGeneId, curGeneChrom, curGeneStart, curGeneEnd);
sr = sqlGetResult(conn, query);
if ((row = sqlNextRow(sr)) != NULL)
{
gp = genePredLoad(row + hasBin);
#define ALIGNIDFIELD 11 // Gencode Id
if (hasAttrId)
curAlignId = cloneString(row[ALIGNIDFIELD]);
else
curAlignId = gp->name;
}
sqlFreeResult(&sr);
if (gp == NULL)
errAbort("getCurGenePred: Can't find %s", query);
return gp;
}
void doFlyreg(struct trackDb *tdb, char *item)
/* flyreg.org: Drosophila DNase I Footprint db. */
{
struct dyString *query = newDyString(256);
struct sqlConnection *conn = hAllocConn(database);
struct sqlResult *sr = NULL;
char **row;
int start = cartInt(cart, "o");
int end = cartInt(cart, "t");
char fullTable[HDB_MAX_TABLE_STRING];
boolean hasBin = FALSE;
char *motifTable = "flyregMotif";
struct dnaMotif *motif = NULL;
boolean isVersion2 = sameString(tdb->table, "flyreg2");
genericHeader(tdb, item);
if (!hFindSplitTable(database, seqName, tdb->table, fullTable, sizeof fullTable, &hasBin))
errAbort("track %s not found", tdb->table);
sqlDyStringPrintf(query, "select * from %s where chrom = '%s' and ",
fullTable, seqName);
hAddBinToQuery(start, end, query);
sqlDyStringPrintf(query, "chromStart = %d and name = '%s'", start, item);
sr = sqlGetResult(conn, query->string);
if ((row = sqlNextRow(sr)) != NULL)
{
struct flyreg2 fr;
if (isVersion2)
flyreg2StaticLoad(row+hasBin, &fr);
else
flyregStaticLoad(row+hasBin, (struct flyreg *)(&fr));
printf("<B>Factor:</B> %s<BR>\n", fr.name);
printf("<B>Target:</B> %s<BR>\n", fr.target);
if (isVersion2)
printf("<B>Footprint ID:</B> %06d<BR>\n", fr.fpid);
printf("<B>PubMed ID:</B> <A HREF=\"");
printEntrezPubMedUidUrl(stdout, fr.pmid);
printf("\" TARGET=_BLANK>%d</A><BR>\n", fr.pmid);
bedPrintPos((struct bed *)(&fr), 3, tdb);
if (hTableExists(database, motifTable))
{
motif = loadDnaMotif(item, motifTable);
if (motif != NULL)
motifHitSection(NULL, motif);
}
}
else
errAbort("query returned no results: \"%s\"", query->string);
dyStringFree(&query);
sqlFreeResult(&sr);
hFreeConn(&conn);
if (motif != NULL)
webNewSection("%s",tdb->longLabel);
printTrackHtml(tdb);
}
struct wiggleDataStream *wigChromRawStats(char *chrom)
/* Fetch stats for wig data in chrom.
* Returns a wiggleDataStream, free it with wiggleDataStreamFree() */
{
char splitTableOrFileName[256];
struct customTrack *ct = NULL;
boolean isCustom = FALSE;
struct wiggleDataStream *wds = NULL;
int operations = wigFetchRawStats;
char *table = curTable;
/* ct, isCustom, wds are set here */
if (isCustomTrack(table))
{
ct = lookupCt(table);
isCustom = TRUE;
if (! ct->wiggle)
errAbort("called to work on a custom track '%s' that isn't wiggle data ?", table);
if (ct->dbTrack)
safef(splitTableOrFileName,ArraySize(splitTableOrFileName), "%s",
ct->dbTableName);
else
safef(splitTableOrFileName,ArraySize(splitTableOrFileName), "%s",
ct->wigFile);
}
wds = wiggleDataStreamNew();
wds->setChromConstraint(wds, chrom);
if (isCustom)
{
if (ct->dbTrack)
wds->getData(wds, CUSTOM_TRASH, splitTableOrFileName, operations);
else
wds->getData(wds, NULL, splitTableOrFileName, operations);
}
else
{
boolean hasBin = FALSE;
if (hFindSplitTable(database, chrom, table, splitTableOrFileName, &hasBin))
{
wds->getData(wds, database, splitTableOrFileName, operations);
}
}
return wds;
}
void chromFeatureSeq(struct sqlConnection *conn,
char *database, char *chrom, char *trackSpec,
FILE *bedFile, FILE *faFile,
int *retItemCount, int *retBaseCount)
/* Write out sequence file for features from one chromosome.
* This separate routine handles the non-merged case. It's
* reason for being is so that the feature names get preserved. */
{
boolean hasBin;
char t[512], *s = NULL;
char table[HDB_MAX_TABLE_STRING];
struct featureBits *fbList = NULL, *fb;
if (trackSpec[0] == '!')
errAbort("Sorry, '!' not available with fa output unless you use faMerge");
isolateTrackPartOfSpec(trackSpec, t);
s = strchr(t, '.');
if (s != NULL)
errAbort("Sorry, only database (not file) tracks allowed with "
"fa output unless you use faMerge");
// ignore isSplit return from hFindSplitTable()
(void) hFindSplitTable(database, chrom, t, table, &hasBin);
fbList = fbGetRangeQuery(database, trackSpec, chrom, 0, hChromSize(database, chrom),
where, TRUE, TRUE);
for (fb = fbList; fb != NULL; fb = fb->next)
{
int s = fb->start, e = fb->end;
if (bedFile != NULL)
{
fprintf(bedFile, "%s\t%d\t%d\t%s",
fb->chrom, fb->start, fb->end, fb->name);
if (fb->strand != '?')
fprintf(bedFile, "\t0\t%c", fb->strand);
fprintf(bedFile, "\n");
}
if (faFile != NULL)
{
struct dnaSeq *seq = hDnaFromSeq(database, chrom, s, e, dnaLower);
if (fb->strand == '-')
reverseComplement(seq->dna, seq->size);
faWriteNext(faFile, fb->name, seq->dna, seq->size);
freeDnaSeq(&seq);
}
}
featureBitsFreeList(&fbList);
}
/* load one or more genePreds from the database */
struct genePred *getPredsForName(char *name, char *geneTable, char *db)
{
struct sqlConnection *conn = hAllocConn(db);
struct genePred *list = NULL;
char splitTable[HDB_MAX_TABLE_STRING];
struct genePred *gene;
boolean hasBin;
struct genePredReader *reader;
boolean found = hFindSplitTable(db, NULL, geneTable,
splitTable, &hasBin);
if (!found)
errAbort("can't find table %s\n", geneTable);
char extra[2048];
if (onlyChrom != NULL)
safef(extra, sizeof extra, "name='%s' and chrom='%s'", name, onlyChrom);
else
safef(extra, sizeof extra, "name='%s'", name);
reader = genePredReaderQuery( conn, splitTable, extra);
while ((gene = genePredReaderNext(reader)) != NULL)
{
verbose(2, "got gene %s\n",gene->name);
slAddHead(&list, gene);
}
if (list == NULL)
errAbort("no genePred for gene %s in %s\n",name, geneTable);
slReverse(&list);
genePredReaderFree(&reader);
hFreeConn(&conn);
return list;
}
static struct psl *loadAlign(struct sqlConnection *conn, struct mappingInfo *mi, int start)
/* load a psl that must exist */
{
char rootTable[256], table[256], query[256];
boolean hasBin;
struct sqlResult *sr;
char **row;
struct psl *psl;
if (mi->suffix == NULL)
safef(rootTable, sizeof(rootTable), "%s%sAli", mi->tblPre, mi->geneSet);
else
safef(rootTable, sizeof(rootTable), "%s%sAli%s", mi->tblPre, mi->geneSet,mi->suffix);
hFindSplitTable(database, seqName, rootTable, table, &hasBin);
sqlSafef(query, sizeof(query), "select * from %s where qName = '%s' and tStart = %d",
table, mi->pg->name, start);
sr = sqlMustGetResult(conn, query);
row = sqlNextRow(sr);
psl = pslLoad(row+hasBin);
sqlFreeResult(&sr);
return psl;
}
struct genePred *getCurGenePred(struct sqlConnection *conn)
/* Return current gene in genePred. */
{
char *track = genomeSetting("knownGene");
char table[64];
boolean hasBin;
char query[256];
struct sqlResult *sr;
char **row;
struct genePred *gp = NULL;
hFindSplitTable(sqlGetDatabase(conn), curGeneChrom, track, table, &hasBin);
sqlSafef(query, sizeof(query),
"select * from %s where name = '%s' "
"and chrom = '%s' and txStart=%d and txEnd=%d"
, table, curGeneId, curGeneChrom, curGeneStart, curGeneEnd);
sr = sqlGetResult(conn, query);
if ((row = sqlNextRow(sr)) != NULL)
gp = genePredLoad(row + hasBin);
sqlFreeResult(&sr);
if (gp == NULL)
errAbort("getCurGenePred: Can't find %s", query);
return gp;
}
struct chain *chainDbLoad(char *db, struct sqlConnection *conn, char *track,
char *chrom, int id)
/** Load chain. */
{
char table[HDB_MAX_TABLE_STRING];
char query[256];
struct sqlResult *sr;
char **row;
int rowOffset;
struct chain *chain = NULL;
if (!hFindSplitTable(db, chrom, track, table, sizeof table, &rowOffset))
errAbort("No %s track in database", track);
sqlSafef(query, sizeof(query),
"select * from %s where id = %d", table, id);
sr = sqlGetResult(conn, query);
row = sqlNextRow(sr);
if (row == NULL)
errAbort("Can't find %d in %s", id, table);
chain = chainHeadLoad(row + rowOffset);
sqlFreeResult(&sr);
chainDbAddBlocks(chain, track, conn);
return chain;
}
static struct psl *loadPslRangeT(char *table, char *qName, char *tName, int tStart, int tEnd)
/* Load a list of psls given qName tName tStart tEnd */
{
struct sqlResult *sr = NULL;
char **row;
struct psl *psl = NULL, *pslList = NULL;
boolean hasBin;
char splitTable[64];
char query[256];
struct sqlConnection *conn = hAllocConn(database);
hFindSplitTable(database, seqName, table, splitTable, &hasBin);
sqlSafef(query, sizeof(query), "select * from %s where qName = '%s' and tName = '%s' and tEnd > %d and tStart < %d", splitTable, qName, tName, tStart, tEnd);
sr = sqlGetResult(conn, query);
while ((row = sqlNextRow(sr)) != NULL)
{
psl = pslLoad(row+hasBin);
slAddHead(&pslList, psl);
}
sqlFreeResult(&sr);
slReverse(&pslList);
hFreeConn(&conn);
return pslList;
}
static int wigOutRegion(char *table, struct sqlConnection *conn,
struct region *region, int maxOut, enum wigOutputType wigOutType,
struct wigAsciiData **data, int spanConstraint)
/* Write out wig data in region. Write up to maxOut elements.
* Returns number of elements written. */
{
int linesOut = 0;
char splitTableOrFileName[HDB_MAX_TABLE_STRING];
struct customTrack *ct = NULL;
boolean isCustom = FALSE;
boolean hasConstraint = FALSE;
struct wiggleDataStream *wds = NULL;
unsigned long long valuesMatched = 0;
int operations = wigFetchAscii;
char *dataConstraint;
double ll = 0.0;
double ul = 0.0;
char *table2 = NULL;
struct bed *intersectBedList = NULL;
switch (wigOutType)
{
case wigOutBed:
operations = wigFetchBed;
break;
default:
case wigDataNoPrint:
case wigOutData:
operations = wigFetchAscii;
break;
};
WIG_INIT; /* ct, isCustom, hasConstraint, wds and table2 are set here */
if (hasConstraint)
freeMem(dataConstraint); /* been cloned into wds */
wds->setMaxOutput(wds, maxOut);
wds->setChromConstraint(wds, region->chrom);
wds->setPositionConstraint(wds, region->start, region->end);
if (table2)
intersectBedList = bedTable2(conn, region, table2);
if (isCustom)
{
if (ct->dbTrack)
{
if (spanConstraint)
wds->setSpanConstraint(wds,spanConstraint);
else
{
struct sqlConnection *trashConn = hAllocConn(CUSTOM_TRASH);
struct trackDb *tdb = findTdbForTable(database, curTrack, table, ctLookupName);
unsigned span = minSpan(trashConn, splitTableOrFileName,
region->chrom, region->start, region->end, cart, tdb);
wds->setSpanConstraint(wds, span);
hFreeConn(&trashConn);
}
valuesMatched = getWigglePossibleIntersection(wds, region,
CUSTOM_TRASH, table2, &intersectBedList,
splitTableOrFileName, operations);
}
else
valuesMatched = getWigglePossibleIntersection(wds, region, NULL, table2,
&intersectBedList, splitTableOrFileName, operations);
}
else
{
if (hFindSplitTable(database, region->chrom, table, splitTableOrFileName, sizeof splitTableOrFileName, NULL))
{
/* XXX TBD, watch for a span limit coming in as an SQL filter */
if (intersectBedList)
{
struct trackDb *tdb = findTdbForTable(database, curTrack, table, ctLookupName);
unsigned span;
span = minSpan(conn, splitTableOrFileName, region->chrom,
region->start, region->end, cart, tdb);
wds->setSpanConstraint(wds, span);
}
else if (spanConstraint)
wds->setSpanConstraint(wds,spanConstraint);
valuesMatched = getWigglePossibleIntersection(wds, region, database,
table2, &intersectBedList, splitTableOrFileName, operations);
}
}
switch (wigOutType)
{
case wigDataNoPrint:
if (data)
{
if (*data != NULL) /* no exercise of this function yet */
{ /* data not null, add to existing list */
struct wigAsciiData *asciiData;
struct wigAsciiData *next;
for (asciiData = *data; asciiData; asciiData = next)
{
next = asciiData->next;
slAddHead(&wds->ascii, asciiData);
}
}
wds->sortResults(wds);
*data = wds->ascii; /* moving the list to *data */
wds->ascii = NULL; /* gone as far as wds is concerned */
}
linesOut = valuesMatched;
break;
case wigOutBed:
linesOut = wds->bedOut(wds, "stdout", TRUE);/* TRUE == sort output */
break;
default:
case wigOutData:
linesOut = wds->asciiOut(wds, database, "stdout", TRUE, FALSE);
break; /* TRUE == sort output, FALSE == not raw data out */
};
wiggleDataStreamFree(&wds);
return linesOut;
} /* static int wigOutRegion() */
void checkInputExists(struct sqlConnection *conn,char *database,
struct chromInfo *chromInfoList, int tableCount, char *tables[])
/* check input tables/files exist, especially to handle split tables */
{
char *track=NULL;
int i = 0, missing=0;
char t[512], *s=NULL;
char table[HDB_MAX_TABLE_STRING];
char fileName[512];
boolean found = FALSE;
for (i=0; i<tableCount; ++i)
{
struct chromInfo *cInfo;
track = tables[i];
if (track[0] == '!')
{
++track;
}
isolateTrackPartOfSpec(track, t);
s = strrchr(t, '.');
if (s)
{
if (fileExists(t))
continue;
}
else
{
if (NULL == conn) conn = hAllocConn(database);
if (sqlTableExists(conn, t))
continue;
}
found = FALSE;
for (cInfo = chromInfoList; cInfo != NULL; cInfo = cInfo->next)
{
if (inclChrom(cInfo->chrom))
{
if (s)
{
chromFileName(t, cInfo->chrom, fileName);
if (fileExists(fileName))
{
found = TRUE;
break;
}
}
else
{
boolean hasBin;
if (hFindSplitTable(database, cInfo->chrom, t, table, &hasBin))
{
found = TRUE;
break;
}
}
}
}
if (!found)
{
if (s)
warn("file %s not found for any chroms", t);
else
warn("table %s not found for any chroms", t);
++missing;
}
}
if (missing>0)
errAbort("Error: %d input table(s)/file(s) do not exist for any of the chroms specified",missing);
}
void doSummaryStatsWiggle(struct sqlConnection *conn)
/* Put up page showing summary stats for wiggle track. */
{
// grab the right trackDb for the current table. The curTrack variable
// has the composite trackDb in it
struct trackDb *track = hTrackDbForTrack(database, curTable);
char *table = curTable;
struct region *region, *regionList = getRegions();
char *regionName = getRegionName();
long long regionSize = 0;
long long gapTotal = 0;
long startTime = 0, wigFetchTime = 0;
char splitTableOrFileName[HDB_MAX_TABLE_STRING];
struct customTrack *ct = NULL;
boolean isCustom = FALSE;
struct wiggleDataStream *wds = NULL;
unsigned long long valuesMatched = 0;
int regionCount = 0;
int regionsDone = 0;
unsigned span = 0;
char *dataConstraint;
double ll = 0.0;
double ul = 0.0;
boolean hasConstraint = FALSE;
char *table2 = NULL;
boolean fullGenome = FALSE;
boolean statsHeaderDone = FALSE;
boolean gotSome = FALSE;
char *shortLabel = table;
long long statsItemCount = 0; /* global accumulators for overall */
int statsSpan = 0; /* stats summary on a multiple region */
double statsSumData = 0.0; /* output */
double statsSumSquares = 0.0; /* " " */
double lowerLimit = INFINITY; /* " " */
double upperLimit = -1.0 * INFINITY; /* " " */
startTime = clock1000();
if (track != NULL)
shortLabel = track->shortLabel;
/* Count the regions, when only one, we can do more stats */
for (region = regionList; region != NULL; region = region->next)
++regionCount;
htmlOpen("%s (%s) Wiggle Summary Statistics", shortLabel, table);
if (anySubtrackMerge(database, curTable))
hPrintf("<P><EM><B>Note:</B> subtrack merge is currently ignored on this "
"page (not implemented yet). Statistics shown here are only for "
"the primary table %s (%s).</EM>", shortLabel, table);
fullGenome = fullGenomeRegion();
WIG_INIT; /* ct, isCustom, hasConstraint, wds and table2 are set here */
for (region = regionList; region != NULL; region = region->next)
{
struct bed *intersectBedList = NULL;
int operations;
++regionsDone;
if (table2)
intersectBedList = bedTable2(conn, region, table2);
operations = wigFetchStats;
#if defined(NOT)
/* can't do the histogram now, that operation times out */
if (1 == regionCount)
operations |= wigFetchAscii;
#endif
wds->setChromConstraint(wds, region->chrom);
if (fullGenome)
wds->setPositionConstraint(wds, 0, 0);
else
wds->setPositionConstraint(wds, region->start, region->end);
if (hasConstraint)
wds->setDataConstraint(wds, dataConstraint, ll, ul);
/* depending on what is coming in on regionList, we may need to be
* smart about how often we call getData for these custom tracks
* since that is potentially a large file read each time.
*/
if (isCustom)
{
if (ct->dbTrack)
{
struct sqlConnection *trashConn = hAllocConn(CUSTOM_TRASH);
struct trackDb *tdb = findTdbForTable(database, curTrack, table, ctLookupName);
span = minSpan(trashConn, splitTableOrFileName, region->chrom,
region->start, region->end, cart, tdb);
wds->setSpanConstraint(wds, span);
valuesMatched = getWigglePossibleIntersection(wds, region,
CUSTOM_TRASH, table2, &intersectBedList,
splitTableOrFileName, operations);
hFreeConn(&trashConn);
}
else
{
valuesMatched = getWigglePossibleIntersection(wds, region, NULL,
table2, &intersectBedList, splitTableOrFileName, operations);
/* XXX We need to properly get the smallest span for custom tracks */
/* This is not necessarily the correct answer here */
if (wds->stats)
span = wds->stats->span;
else
span = 1;
}
}
else
{
if (hFindSplitTable(database, region->chrom, table, splitTableOrFileName, sizeof splitTableOrFileName, NULL))
{
span = minSpan(conn, splitTableOrFileName, region->chrom,
region->start, region->end, cart, track);
wds->setSpanConstraint(wds, span);
valuesMatched = getWigglePossibleIntersection(wds, region,
database, table2, &intersectBedList, splitTableOrFileName,
operations);
if (intersectBedList)
span = 1;
}
}
/* when doing multiple regions, we need to print out each result as
* it happens to keep the connection open to the browser and
* prevent any timeout since this could take a while.
* (worst case test is quality track on panTro1)
*/
if (wds->stats)
statsItemCount += wds->stats->count;
if (wds->stats && (regionCount > 1) && (valuesMatched > 0))
{
double sumData = wds->stats->mean * wds->stats->count;
double sumSquares;
if (wds->stats->count > 1)
sumSquares = (wds->stats->variance * (wds->stats->count - 1)) +
((sumData * sumData)/wds->stats->count);
else
sumSquares = sumData * sumData;
/* global accumulators for overall summary */
statsSpan = wds->stats->span;
statsSumData += sumData;
statsSumSquares += sumSquares;
if (wds->stats->lowerLimit < lowerLimit)
lowerLimit = wds->stats->lowerLimit;
if ((wds->stats->lowerLimit + wds->stats->dataRange) > upperLimit)
upperLimit = wds->stats->lowerLimit + wds->stats->dataRange;
if (statsHeaderDone)
wds->statsOut(wds, database, "stdout", TRUE, TRUE, FALSE, TRUE);
else
{
wds->statsOut(wds, database, "stdout", TRUE, TRUE, TRUE, TRUE);
statsHeaderDone = TRUE;
}
wds->freeStats(wds);
gotSome = TRUE;
}
if ((regionCount > MAX_REGION_DISPLAY) &&
(regionsDone >= MAX_REGION_DISPLAY))
{
hPrintf("<TR><TH ALIGN=CENTER COLSPAN=12> Can not display more "
"than %d regions, <BR> would take too much time </TH></TR>\n",
MAX_REGION_DISPLAY);
break; /* exit this for loop */
}
} /*for (region = regionList; region != NULL; region = region->next) */
if (hasConstraint)
freeMem(dataConstraint); /* been cloned into wds */
if (1 == regionCount)
{
statsPreamble(wds, regionList->chrom, regionList->start, regionList->end,
span, valuesMatched, table2);
/* 3 X TRUE = sort results, html table output, with header,
* the FALSE means close the table after printing, no more rows to
* come. The case in the if() statement was already taken care of
* in the statsPreamble() printout. No need to do that again.
*/
if ( ! ((valuesMatched == 0) && table2) )
wds->statsOut(wds, database, "stdout", TRUE, TRUE, TRUE, FALSE);
regionSize = basesInRegion(regionList,0);
gapTotal = gapsInRegion(conn, regionList,0);
}
else
{ /* this is a bit of a kludge here since these printouts are done in the
* library source wigDataStream.c statsOut() function and
* this is a clean up of that. That function should be
* pulled out of there and made independent and more
* versatile.
*/
long long realSize;
double variance;
double stddev;
/* Too expensive to lookup the numbers for thousands of regions */
regionSize = basesInRegion(regionList,MAX_REGION_DISPLAY);
gapTotal = gapsInRegion(conn, regionList,MAX_REGION_DISPLAY);
realSize = regionSize - gapTotal;
/* close the table which was left open in the loop above */
if (!gotSome)
hPrintf("<TR><TH ALIGN=CENTER COLSPAN=12> No data found matching this request </TH></TR>\n");
hPrintf("<TR><TH ALIGN=LEFT> SUMMARY: </TH>\n");
hPrintf("\t<TD> </TD>\n"); /* chromStart */
hPrintf("\t<TD> </TD>\n"); /* chromEnd */
hPrintf("\t<TD ALIGN=RIGHT> ");
printLongWithCommas(stdout, statsItemCount);
hPrintf(" </TD>\n" );
hPrintf("\t<TD ALIGN=RIGHT> %d </TD>\n", statsSpan);
hPrintf("\t<TD ALIGN=RIGHT> ");
printLongWithCommas(stdout, statsItemCount*statsSpan);
hPrintf(" (%.2f%%) </TD>\n",
100.0*(double)(statsItemCount*statsSpan)/(double)realSize);
hPrintf("\t<TD ALIGN=RIGHT> %g </TD>\n", lowerLimit);
hPrintf("\t<TD ALIGN=RIGHT> %g </TD>\n", upperLimit);
hPrintf("\t<TD ALIGN=RIGHT> %g </TD>\n", upperLimit - lowerLimit);
if (statsItemCount > 0)
hPrintf("\t<TD ALIGN=RIGHT> %g </TD>\n", statsSumData/statsItemCount);
else
hPrintf("\t<TD ALIGN=RIGHT> 0.0 </TD>\n");
stddev = 0.0;
variance = 0.0;
if (statsItemCount > 1)
{
variance = (statsSumSquares -
((statsSumData * statsSumData)/(double) statsItemCount)) /
(double) (statsItemCount - 1);
if (variance > 0.0)
stddev = sqrt(variance);
}
hPrintf("\t<TD ALIGN=RIGHT> %g </TD>\n", variance);
hPrintf("\t<TD ALIGN=RIGHT> %g </TD>\n", stddev);
hPrintf("</TR>\n");
wigStatsTableHeading(stdout, TRUE);
hPrintf("</TABLE></TD></TR></TABLE></P>\n");
}
#if defined(NOT)
/* can't do the histogram now, that operation times out */
/* Single region, we can do the histogram */
if ((valuesMatched > 1) && (1 == regionCount))
{
float *valuesArray = NULL;
size_t valueCount = 0;
struct histoResult *histoGramResult;
/* convert the ascii data listings to one giant float array */
valuesArray = wds->asciiToDataArray(wds, valuesMatched, &valueCount);
/* histoGram() may return NULL if it doesn't work */
histoGramResult = histoGram(valuesArray, valueCount,
NAN, (unsigned) 0, NAN, (float) wds->stats->lowerLimit,
(float) (wds->stats->lowerLimit + wds->stats->dataRange),
(struct histoResult *)NULL);
printHistoGram(histoGramResult, TRUE); /* TRUE == html output */
freeHistoGram(&histoGramResult);
wds->freeAscii(wds);
wds->freeArray(wds);
}
#endif
wds->freeStats(wds);
wiggleDataStreamFree(&wds);
wigFetchTime = clock1000() - startTime;
webNewSection("Region and Timing Statistics");
hTableStart();
stringStatRow("region", regionName);
numberStatRow("bases in region", regionSize);
numberStatRow("bases in gaps", gapTotal);
floatStatRow("load and calc time", 0.001*wigFetchTime);
wigFilterStatRow(conn);
stringStatRow("intersection", cartUsualString(cart, hgtaIntersectTable, "off"));
hTableEnd();
htmlClose();
} /* void doSummaryStatsWiggle(struct sqlConnection *conn) */
struct bed *getWiggleAsBed(
char *db, char *table, /* Database and table. */
struct region *region, /* Region to get data for. */
char *filter, /* Filter to add to SQL where clause if any. */
struct hash *idHash, /* Restrict to id's in this hash if non-NULL. */
struct lm *lm, /* Where to allocate memory. */
struct sqlConnection *conn) /* SQL connection to work with */
/* Return a bed list of all items in the given range in table.
* Cleanup result via lmCleanup(&lm) rather than bedFreeList. */
/* filter, idHash and lm are currently unused, perhaps future use */
{
struct bed *bedList=NULL;
char splitTableOrFileName[HDB_MAX_TABLE_STRING];
struct customTrack *ct = NULL;
boolean isCustom = FALSE;
boolean hasConstraint = FALSE;
struct wiggleDataStream *wds = NULL;
unsigned long long valuesMatched = 0;
int operations = wigFetchBed;
char *dataConstraint;
double ll = 0.0;
double ul = 0.0;
char *table2 = NULL;
struct bed *intersectBedList = NULL;
int maxOut;
WIG_INIT; /* ct, isCustom, hasConstraint, wds and table2 are set here */
if (hasConstraint)
freeMem(dataConstraint); /* been cloned into wds */
maxOut = bigFileMaxOutput();
wds->setMaxOutput(wds, maxOut);
wds->setChromConstraint(wds, region->chrom);
wds->setPositionConstraint(wds, region->start, region->end);
if (table2)
intersectBedList = bedTable2(conn, region, table2);
if (isCustom)
{
if (ct->dbTrack)
{
unsigned span = 0;
struct sqlConnection *trashConn = hAllocConn(CUSTOM_TRASH);
struct trackDb *tdb = findTdbForTable(database, curTrack, table, ctLookupName);
valuesMatched = getWigglePossibleIntersection(wds, region,
CUSTOM_TRASH, table2, &intersectBedList,
splitTableOrFileName, operations);
span = minSpan(trashConn, splitTableOrFileName, region->chrom,
region->start, region->end, cart, tdb);
wds->setSpanConstraint(wds, span);
hFreeConn(&trashConn);
}
else
valuesMatched = getWigglePossibleIntersection(wds, region, NULL, table2,
&intersectBedList, splitTableOrFileName, operations);
}
else
{
if (conn == NULL)
errAbort( "getWiggleAsBed: NULL conn given for database table");
if (hFindSplitTable(database, region->chrom, table, splitTableOrFileName, sizeof splitTableOrFileName, NULL))
{
struct trackDb *tdb = findTdbForTable(database, curTrack, table, ctLookupName);
unsigned span = 0;
/* XXX TBD, watch for a span limit coming in as an SQL filter */
span = minSpan(conn, splitTableOrFileName, region->chrom,
region->start, region->end, cart, tdb);
wds->setSpanConstraint(wds, span);
valuesMatched = getWigglePossibleIntersection(wds, region, database,
table2, &intersectBedList, splitTableOrFileName, operations);
}
}
if (valuesMatched > 0)
{
struct bed *bed;
wds->sortResults(wds);
for (bed = wds->bed; bed != NULL; bed = bed->next)
{
struct bed *copy = lmCloneBed(bed, lm);
slAddHead(&bedList, copy);
}
slReverse(&bedList);
}
wiggleDataStreamFree(&wds);
return bedList;
} /* struct bed *getWiggleAsBed() */
void genericWiggleClick(struct sqlConnection *conn, struct trackDb *tdb,
char *item, int start)
/* Display details for Wiggle data tracks.
* conn may be NULL for custom tracks when from file */
{
char *chrom = cartString(cart, "c");
char table[64];
boolean hasBin;
unsigned span = 0;
struct wiggleDataStream *wds = wiggleDataStreamNew();
unsigned long long valuesMatched = 0;
struct histoResult *histoGramResult;
float *valuesArray = NULL;
size_t valueCount = 0;
struct customTrack *ct = NULL;
boolean isCustom = FALSE;
int operations = wigFetchStats; /* default operation */
if (startsWith("ct_", tdb->table))
{
ct = lookupCt(tdb->table);
if (!ct)
{
warn("<P>wiggleClick: can not find custom wiggle track '%s'</P>", tdb->table);
return;
}
if (! ct->wiggle)
{
warn("<P>wiggleClick: called to do stats on a custom track that isn't wiggle data ?</P>");
return;
}
if (ct->dbTrack)
{
safef(table,ArraySize(table), "%s", ct->dbTableName);
span = minSpan(conn, table, chrom, winStart, winEnd, cart, tdb);
}
else
{
safef(table,ArraySize(table), "%s", ct->wigFile);
span = 0; /* cause all spans to be examined */
}
isCustom = TRUE;
}
else
{
hFindSplitTable(database, seqName, tdb->table, table, &hasBin);
/*span = spanInUse(conn, table, chrom, winStart, winEnd, cart);*/
span = minSpan(conn, table, chrom, winStart, winEnd, cart, tdb);
}
/* if for some reason we don't have a chrom and win positions, this
* should be run in a loop that does one chrom at a time. In the
* case of hgc, there seems to be a chrom and a position.
*/
wds->setSpanConstraint(wds, span);
wds->setChromConstraint(wds, chrom);
wds->setPositionConstraint(wds, winStart, winEnd);
/* If our window is less than some number of points, we can do
* the histogram too.
*/
#define MAX_WINDOW_ALLOW_STATS 100000001
#define MAX_WINDOW_ALLOW_STRING "100,000,000"
if ((winEnd - winStart) < MAX_WINDOW_ALLOW_STATS)
operations |= wigFetchAscii;
/* We want to also fetch the actual data values so we can run a
* histogram function on them. You can't fetch the data in the
* form of the data array since the span information is then lost.
* We have to do the ascii data list format, and prepare that to
* send to the histogram function.
*/
if (isCustom)
{
if (ct->dbTrack)
valuesMatched = wds->getData(wds, CUSTOM_TRASH, table, operations);
else
valuesMatched = wds->getData(wds, (char *)NULL, table, operations);
}
else
valuesMatched = wds->getData(wds, database, table, operations);
statsPreamble(wds, chrom, winStart, winEnd, span, valuesMatched, NULL);
/* output statistics table
* (+sort, +html output, +with header, +close table)
*/
wds->statsOut(wds, database, "stdout", TRUE, TRUE, TRUE, FALSE);
if ((winEnd - winStart) < MAX_WINDOW_ALLOW_STATS)
{
char *words[16];
int wordCount = 0;
char *dupe = cloneString(tdb->type);
double minY, maxY, tDbMinY, tDbMaxY;
float hMin, hMax, hRange;
wordCount = chopLine(dupe, words);
wigFetchMinMaxY(tdb, &minY, &maxY, &tDbMinY, &tDbMaxY, wordCount, words);
hMin = min(minY,tDbMinY);
hMax = max(maxY,tDbMaxY);
hRange = hMax - hMin;
/* convert the ascii data listings to one giant float array */
valuesArray = wds->asciiToDataArray(wds, valuesMatched, &valueCount);
/* let's see if we really want to use the range from the track type
* line, or the actual range in this data. If there is a good
* actual range in the data, use that instead
*/
if (hRange > 0.0)
{
if (wds->stats->dataRange != 0)
hRange = 0.0;
}
/* If we have a valid range, use a specified 20 bin histogram
* NOTE: pass 21 as binCount to get a 20 bin histogram
*/
if (hRange > 0.0)
histoGramResult = histoGram(valuesArray, valueCount, (hRange/20.0),
(unsigned) 21, hMin, hMin, hMax, (struct histoResult *)NULL);
else
histoGramResult = histoGram(valuesArray, valueCount,
NAN, (unsigned) 0, NAN, (float) wds->stats->lowerLimit,
(float) (wds->stats->lowerLimit + wds->stats->dataRange),
(struct histoResult *)NULL);
/* histoGram() may return NULL if it doesn't work, that's OK, the
* print out will indicate no results (TRUE == html output)
*/
printHistoGram(histoGramResult, TRUE);
freeHistoGram(&histoGramResult);
freeMem(valuesArray);
}
else
{
printf("<P>(viewing windows of fewer than %s bases will also"
" display a histogram)</P>\n", MAX_WINDOW_ALLOW_STRING);
}
wiggleDataStreamFree(&wds);
}
