static void selectFromSplitTable(char *db, char *table, struct hTableInfo *tblInfo,
struct sqlConnection *conn, FILE *outFh)
/* select from a split table */
{
struct slName *chroms = hAllChromNames(db);
struct slName *chrom;
char chromTable[256];
for (chrom = chroms; chrom != NULL ; chrom = chrom->next)
if (inclChrom(chrom->name))
{
safef(chromTable, sizeof(chromTable), "%s_%s", chrom->name, table);
selectFromTable(chromTable, tblInfo, conn, outFh);
}
}
void hgChroms(char *db)
/* hgChroms - print chromosomes for a genome. */
{
struct slName *chrom, *chroms = hAllChromNames(db);
for (chrom = chroms; chrom != NULL; chrom = chrom->next)
{
if (inclChrom(chrom))
{
if (noPrefix && startsWith("chr", chrom->name))
{
printf("%s\n", chrom->name + strlen("chr"));
}
else
{
printf("%s\n", chrom->name);
}
}
}
}
void featureBits(char *database, int tableCount, char *tables[])
/* featureBits - Correlate tables via bitmap projections and booleans. */
{
struct sqlConnection *conn = NULL;
char *bedName = optionVal("bed", NULL), *faName = optionVal("fa", NULL);
char *binName = optionVal("bin", NULL);
char *bedRegionInName = optionVal("bedRegionIn", NULL);
char *bedRegionOutName = optionVal("bedRegionOut", NULL);
FILE *bedFile = NULL, *faFile = NULL, *binFile = NULL;
FILE *bedRegionOutFile = NULL;
struct bed *bedRegionList = NULL;
boolean faIndependent = FALSE;
struct chromInfo *cInfo;
if (bedName)
bedFile = mustOpen(bedName, "w");
if (binName)
binFile = mustOpen(binName, "w");
if ((bedRegionInName && !bedRegionOutName) || (!bedRegionInName && bedRegionOutName))
errAbort("bedRegionIn and bedRegionOut must both be specified");
if (faName)
{
boolean faMerge = optionExists("faMerge");
faFile = mustOpen(faName, "w");
if (tableCount > 1)
{
if (!faMerge)
errAbort("For fa output of multiple tables you must use the "
"faMerge option");
}
faIndependent = (!faMerge);
}
if (chromSizes != NULL)
chromInfoList = chromInfoLoadAll(chromSizes);
else
chromInfoList = fbCreateChromInfoList(clChrom, database);
if (!countGaps)
conn = hAllocConn(database);
checkInputExists(conn, database, chromInfoList, tableCount, tables);
if (!faIndependent)
{
double totalBases = 0, totalBits = 0;
int firstTableBits = 0, secondTableBits = 0;
int *pFirstTableBits = NULL, *pSecondTableBits = NULL;
double totalFirstBits = 0, totalSecondBits = 0;
static int dotClock = 1;
if (calcEnrichment)
{
pFirstTableBits = &firstTableBits;
pSecondTableBits = &secondTableBits;
}
if (bedRegionInName)
{
struct lineFile *lf = lineFileOpen(bedRegionInName, TRUE);
struct bed *bed;
char *row[3];
bedRegionOutFile = mustOpen(bedRegionOutName, "w");
while (lineFileRow(lf, row))
{
if (startsWith(row[0],"#")||startsWith(row[0],"chrom"))
continue;
bed = bedLoad3(row);
slAddHead(&bedRegionList, bed);
}
lineFileClose(&lf);
slReverse(&bedRegionList);
}
for (cInfo = chromInfoList; cInfo != NULL; cInfo = cInfo->next)
{
if (inclChrom(cInfo->chrom))
{
int chromBitSize;
int chromSize = cInfo->size;
verbose(3,"chromFeatureBits(%s)\n", cInfo->chrom);
chromFeatureBits(conn, database, cInfo->chrom, tableCount, tables,
bedFile, faFile, binFile, bedRegionList, bedRegionOutFile,
chromSize, &chromBitSize, pFirstTableBits, pSecondTableBits
);
totalBases += countBases(conn, cInfo->chrom, chromSize, database);
totalBits += chromBitSize;
totalFirstBits += firstTableBits;
totalSecondBits += secondTableBits;
if (dots > 0)
{
if (--dotClock <= 0)
{
fputc('.', stdout);
fflush(stdout);
dotClock = dots;
}
}
}
}
if (dots > 0)
{
fputc('\n', stdout);
fflush(stdout);
}
if (calcEnrichment)
fprintf(stderr,"%s %5.3f%%, %s %5.3f%%, both %5.3f%%, cover %4.2f%%, enrich %4.2fx\n",
tables[0],
100.0 * totalFirstBits/totalBases,
tables[1],
100.0 * totalSecondBits/totalBases,
100.0 * totalBits/totalBases,
100.0 * totalBits / totalFirstBits,
(totalBits/totalSecondBits) / (totalFirstBits/totalBases) );
else
fprintf(stderr,"%1.0f bases of %1.0f (%4.3f%%) in intersection\n",
totalBits, totalBases, 100.0*totalBits/totalBases);
}
else
{
int totalItems = 0;
double totalBases = 0;
int itemCount, baseCount;
for (cInfo = chromInfoList; cInfo != NULL; cInfo = cInfo->next)
{
if (inclChrom(cInfo->chrom))
{
chromFeatureSeq(conn, database, cInfo->chrom, tables[0],
bedFile, faFile, &itemCount, &baseCount);
totalBases += countBases(conn, cInfo->chrom, baseCount, database);
totalItems += itemCount;
}
}
}
hFreeConn(&conn);
}
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);
}
