struct hash *loadGenePositions(char *database, struct sqlConnection *conn, char *fileName)
/* Read in 7 column file and convert to hash of gene
* positions. */
{
struct hash *hash = newHash(16);
struct lineFile *lf = lineFileOpen(fileName, TRUE);
char *row[7];
int count = 0;
struct genomePos *posList = NULL, *pos;
while (lineFileRow(lf, row))
{
int geneStart,geneEnd,upSize,downSize;
AllocVar(pos);
hashAddSaveName(hash, row[0], pos, &pos->name);
slAddHead(&posList, pos);
pos->chrom = hgOfficialChromName(database, row[1]);
if (pos->chrom == NULL)
errAbort("Unrecognized chromosome %s line %d of %s",
row[1], lf->lineIx, lf->fileName);
geneStart = lineFileNeedNum(lf, row, 2);
geneEnd = lineFileNeedNum(lf, row, 3);
pos->strand = row[4][0];
if (pos->strand != '+' && pos->strand != '-')
errAbort("Unrecognized strand %s line %d of %s",
row[4], lf->lineIx, lf->fileName);
upSize = lineFileNeedNum(lf, row, 5);
downSize = lineFileNeedNum(lf, row, 6);
if (pos->strand == '+')
{
pos->start = geneStart - upSize;
pos->end = geneStart + downSize;
}
else
{
pos->start = geneEnd - downSize;
pos->end = geneEnd + upSize;
}
++count;
}
verbose(1, "%d genes in %s\n", count, fileName);
return hash;
}
void hgDeleteChrom(char *db, int chromCount, char *chromNames[])
/* hgDeleteChrom - output SQL commands to delete chrom(s) from db. */
{
struct sqlConnection *conn = hAllocConn(db);
struct sqlResult *sr = NULL;
char **row = NULL;
int i;
for (i=0; i < chromCount; i++)
{
char *chrom = hgOfficialChromName(db, chromNames[i]);
if (chrom == NULL)
errAbort("Error: \"%s\" is not a chromosome in %s.",
chromNames[i], db);
printf("delete from %s.chromInfo where chrom = \"%s\";\n", db, chrom);
sr = sqlGetResult(conn, "NOSQLINJ SHOW TABLES");
while((row = sqlNextRow(sr)) != NULL)
{
char *table = row[0];
char tChrom[32];
char rootName[128];
struct hTableInfo *hti = NULL;
if (sscanf(table, "chr%32[^_]_random_%128s", tChrom, rootName) == 2 ||
sscanf(table, "chr%32[^_]_%128s", tChrom, rootName) == 2)
hti = hFindTableInfo(db, chrom, rootName);
else
hti = hFindTableInfo(db, chrom, table);;
if (hti->isPos)
{
if (hti->isSplit)
{
char tableCmp[256];
safef(tableCmp, sizeof(tableCmp), "%s_%s", chrom, hti->rootName);
if (sameWord(table, tableCmp))
printf("drop table %s.%s;\n", db, table);
}
else
printf("delete from %s.%s where %s = \"%s\";\n",
db, table, hti->chromField, chrom);
}
}
sqlFreeResult(&sr);
}
hFreeConn(&conn);
}
int main(int argc, char *argv[])
/* Check args and call snpMaskGenes. */
{
if (argc != 5)
usage();
database = argv[1];
if(!hDbExists(database))
errAbort("%s does not exist\n", database);
hSetDb(database);
if(!hTableExistsDb(database, "snp"))
errAbort("no snp table in %s\n", database);
chromName = argv[2];
if(hgOfficialChromName(chromName) == NULL)
errAbort("no such chromosome %s in %s\n", chromName, database);
// check that nib file exists
// or, use hNibForChrom from hdb.c
snpMaskGenes(argv[3], argv[4]);
return 0;
}
static struct bed4 *parseRegionInput(char *db, char *inputString, int maxRegions, int maxErrs,
struct dyString *dyWarn)
/* scan the user region definition, turn into a bed list */
{
int regionCount = 0;
int errCount = 0;
struct bed4 *bedList = NULL;
struct lineFile *lf = lineFileOnString("userData", TRUE, inputString);
char *line = NULL;
while (lineFileNextReal(lf, &line))
{
char *chromName = NULL;
int chromStart = 0;
int chromEnd = 0;
char *regionName = NULL;
// Chop a copy of line so we can display line if there's an error.
char copy[strlen(line)+1];
safecpy(copy, sizeof(copy), line);
char *words[5];
int wordCount = chopByWhite(copy, words, ArraySize(words));
boolean badFormat = FALSE;
boolean gotError = FALSE;
/* might be something of the form: chrom:start-end optionalRegionName */
if (((1 == wordCount) || (2 == wordCount)) &&
hgParseChromRange(NULL, words[0], &chromName,
&chromStart, &chromEnd))
{
if (2 == wordCount)
regionName = cloneString(words[1]);
}
else if (!((3 == wordCount) || (4 == wordCount)))
{
dyStringPrintf(dyWarn, "line %d: '%s': "
"unrecognized format. Please enter 3- or 4-column BED or "
"a chr:start-end position range optionally followed by a name.\n",
lf->lineIx, line);
badFormat = TRUE;
gotError = TRUE;
}
else
{
chromName = words[0];
// Make sure chromStart and chromEnd are numbers
if (!isNumericString(words[1]))
{
dyStringPrintf(dyWarn, "line %d: '%s': chromStart must be a number but is '%s'\n",
lf->lineIx, line, words[1]);
gotError = TRUE;
}
if (!isNumericString(words[2]))
{
dyStringPrintf(dyWarn, "line %d: '%s': chromEnd must be a number but is '%s'\n",
lf->lineIx, line, words[2]);
gotError = TRUE;
}
if (! gotError)
{
chromStart = atoi(words[1]);
chromEnd = atoi(words[2]);
if (wordCount > 3)
regionName = cloneString(words[3]);
}
}
char *officialChromName = chromName ? hgOfficialChromName(db, chromName) : NULL;
if (! badFormat)
{
if (NULL == officialChromName)
{
dyStringPrintf(dyWarn,
"line %d: '%s': chrom name '%s' not recognized in this assembly\n",
lf->lineIx, line, chromName ? chromName : words[0]);
gotError = TRUE;
}
else if (illegalCoordinate(db, officialChromName, chromStart, chromEnd, line, lf->lineIx,
dyWarn))
{
gotError = TRUE;
}
}
if (gotError)
{
errCount++;
if (errCount > maxErrs && maxErrs > 0)
{
dyStringPrintf(dyWarn, "Exceeded maximum number of errors (%d), quitting\n", maxErrs);
break;
}
else
continue;
}
++regionCount;
if (regionCount > maxRegions && maxRegions > 0)
{
dyStringPrintf(dyWarn,
"line %d: limit of %d region definitions exceeded, skipping the rest\n",
lf->lineIx, maxRegions);
break;
}
struct bed4 *bedEl = bed4New(officialChromName, chromStart, chromEnd, regionName);
slAddHead(&bedList, bedEl);
}
lineFileClose(&lf);
// Keep regions in same order as user entered them:
slReverse(&bedList);
return (bedList);
}
static struct bed *parseRegionInput(char *inputString)
/* scan the user region definition, turn into a bed list */
{
int itemCount = 0;
struct bed *bedList = NULL;
struct bed *bedEl;
int wordCount;
char *words[5];
struct lineFile *lf;
lf = lineFileOnString("userData", TRUE, inputString);
while (0 != (wordCount = lineFileChopNext(lf, words, ArraySize(words))))
{
char *chromName = NULL;
int chromStart = 0;
int chromEnd = 0;
char *regionName = NULL;
/* might be something of the form: chrom:start-end optionalRegionName */
if (((1 == wordCount) || (2 == wordCount)) &&
hgParseChromRange(NULL, words[0], &chromName,
&chromStart, &chromEnd))
{
if (2 == wordCount)
regionName = cloneString(words[1]);
}
else if (!((3 == wordCount) || (4 == wordCount)))
{
int i;
struct dyString *errMessage = dyStringNew(0);
for (i = 0; i < wordCount; ++i)
dyStringPrintf(errMessage, "%s ", words[i]);
errAbort("line %d: '%s'<BR>\n"
"illegal bed size, expected 3 or 4 fields, found %d\n",
lf->lineIx, dyStringCannibalize(&errMessage), wordCount);
}
else
{
chromName = hgOfficialChromName(database, words[0]);
chromStart = sqlSigned(words[1]);
chromEnd = sqlSigned(words[2]);
if (wordCount > 3)
regionName = cloneString(words[3]);
}
++itemCount;
if (itemCount > 1000)
{
warn("limit 1000 region definitions reached at line %d<BR>\n",
lf->lineIx);
break;
}
AllocVar(bedEl);
bedEl->chrom = chromName;
if (NULL == bedEl->chrom)
errAbort("at line %d, chrom name '%s' %s %s not recognized in this assembly %d",
lf->lineIx, words[0], words[1], words[2], wordCount);
bedEl->chromStart = chromStart;
bedEl->chromEnd = chromEnd;
if (illegalCoordinate(bedEl->chrom, bedEl->chromStart, bedEl->chromEnd))
errAbort("illegal input at line %d: %s %d %d",
lf->lineIx, bedEl->chrom, bedEl->chromStart, bedEl->chromEnd);
if (wordCount > 3)
bedEl->name = regionName;
else
bedEl->name = NULL;
/* if we wanted to give artifical names to each item */
#ifdef NOT
{
char name[128];
safef(name, ArraySize(name), "item_%04d", itemCount);
bedEl->name = cloneString(name);
}
#endif
slAddHead(&bedList, bedEl);
}
lineFileClose(&lf);
// slSort(&bedList, bedCmp); /* this would do chrom,chromStart order */
slReverse(&bedList); /* with no sort, it is in order as user entered */
return (bedList);
}
