struct sqlConnection *sqlWarnConnect(char *db)
/* Connect to database, or warn and return NULL. */
{
struct sqlConnection *conn = sqlMayConnect(db);
if (conn == NULL)
warn("Error: Couldn't connect to database %s", db);
return conn;
}
static boolean isInMBLabValidDb(char *acc)
/* check if an accession is in the Brent lab validation database */
{
boolean inMBLabValidDb = FALSE;
struct sqlConnection *fconn = sqlMayConnect("hgFixed");
if ((fconn != NULL) && sqlTableExists(fconn, "mgcMBLabValid"))
{
char query[64], buf[32];
sqlSafef(query, sizeof(query), "select acc from mgcMBLabValid where acc=\"%s\"",
acc);
if (sqlQuickQuery(fconn, query, buf, sizeof(buf)) != NULL)
inMBLabValidDb = TRUE;
sqlDisconnect(&fconn);
}
return inMBLabValidDb;
}
void joinerCheckTableCoverage(struct joiner *joiner, char *specificDb)
/* Check that all tables either are part of an identifier or
* are in the tablesIgnored statements. */
{
struct slName *miss, *missList = NULL;
struct hashEl *dbList, *db;
dbList = hashElListHash(joiner->databasesChecked);
for (db = dbList; db != NULL; db = db->next)
{
if (specificDb == NULL || sameString(db->name, specificDb))
{
struct sqlConnection *conn = sqlMayConnect(db->name);
if (conn == NULL)
warn("Error: database %s doesn't exist", db->name);
else
{
struct slName *table;
struct slName *tableList = sqlListTables(conn);
struct hash *hash = getCoveredTables(joiner, db->name, conn);
for (table = tableList; table != NULL; table = table->next)
{
if (!hashLookup(hash, table->name))
{
char fullName[256];
safef(fullName, sizeof(fullName), "%s.%s",
db->name, table->name);
miss = slNameNew(fullName);
slAddHead(&missList, miss);
}
else
verbose(2,"tableCovered: '%s'\n", table->name);
}
slFreeList(&tableList);
freeHash(&hash);
reportErrorList(&missList, "tables not in .joiner file");
}
sqlDisconnect(&conn);
}
}
slFreeList(&dbList);
}
char *uniProtFindPrimAcc(char *id)
/* Return primary accession given an alias. */
/* The alias could be an accession number, display ID, old display ID, etc.
* NULL if not found. */
{
static struct sqlConnection *conn=NULL;
char *acc;
char query[256];
if (conn==NULL)
{
conn = sqlMayConnect(PROTEOME_DB_NAME);
if (conn == NULL) return NULL;
}
safef(query, sizeof(query), "select acc from uniProtAlias where alias = '%s'", id);
acc = sqlQuickString(conn, query);
return(acc);
}
void joinerCheckDependencies(struct joiner *joiner, char *specificDb)
/* Check time stamps on dependent tables. */
{
struct hashEl *db, *dbList = hashElListHash(joiner->databasesChecked);
for (db = dbList; db != NULL; db = db->next)
{
if (specificDb == NULL || sameString(specificDb, db->name))
{
struct sqlConnection *conn = sqlMayConnect(db->name);
if (conn != NULL) /* We've already warned on this NULL */
{
struct joinerDependency *dep;
for (dep = joiner->dependencyList; dep != NULL; dep = dep->next)
{
checkOneDependency(joiner, dep, conn, db->name);
}
sqlDisconnect(&conn);
}
}
}
slFreeList(&dbList);
}
char *uniProtFindPrimAccFromGene(char *gene, char *db)
/* Return primary accession given gene name.
* NULL if not found. */
{
static struct sqlConnection *conn=NULL;
char *acc = NULL;
char query[256];
char *pdb = cloneString(UNIPROT_DB_NAME);
char *geneTable = "gene";
if (conn==NULL)
{
conn = sqlMayConnect(pdb);
if (conn == NULL) return NULL;
}
if (geneTable != NULL && sqlTableExists(conn,geneTable))
{
sqlSafef(query, sizeof(query), "select g.acc from %s g , accToTaxon a, taxon t where val = '%s' and g.acc = a.acc and id = a.taxon and binomial = '%s' ",geneTable, gene, hGenome(db));
acc = sqlQuickString(conn, query);
}
return(acc);
}
void doKeyChecks(char *db, struct joiner *joiner, struct joinerSet *js,
struct hash *keyHash, struct keyHitInfo *khiList,
struct joinerField *keyField, struct joinerField *jf)
/* Check that at least minimum is covered, and that full and
* unique tags are respected. */
{
struct sqlConnection *conn = sqlMayConnect(db);
if (conn != NULL)
{
boolean okFlag = FALSE;
int total = 0, hits = 0, hitsNeeded;
char *miss = NULL;
struct slName *table;
struct slName *tableList = getTablesForField(conn,jf->splitPrefix,
jf->table, jf->splitSuffix);
keyHitInfoClear(khiList);
for (table = tableList; table != NULL; table = table->next)
{
char query[256], **row;
struct sqlResult *sr;
safef(query, sizeof(query), "select %s from %s",
jf->field, table->name);
sr = sqlGetResult(conn, query);
while ((row = sqlNextRow(sr)) != NULL)
{
if (jf->separator == NULL)
{
addHitMiss(keyHash, row[0], jf, &hits, &miss, &total);
}
else
{
/* Do list. */
struct slName *el, *list;
int ix;
list = slNameListFromString(row[0], jf->separator[0]);
for (el = list, ix=0; el != NULL; el = el->next, ++ix)
{
char *id = el->name;
char buf[16];
if (jf->indexOf)
{
safef(buf, sizeof(buf), "%d", ix);
id = buf;
}
addHitMiss(keyHash, id, jf, &hits, &miss, &total);
}
slFreeList(&list);
}
}
sqlFreeResult(&sr);
}
if (tableList != NULL)
{
okFlag = FALSE;
if (hits==total) okFlag = TRUE;
if (jf->minCheck < 0.999999) /* Control for rounding error */
hitsNeeded = round(total * jf->minCheck);
else
hitsNeeded = total;
if (jf->minCheck < 1.0 && hits >= hitsNeeded) okFlag = TRUE;
verbose(1, " %s.%s.%s - hits %d of %d%s\n", db, jf->table, jf->field, hits, total,
okFlag ? " ok" : js->isFuzzy ? " fuzzy" : "");
if (hits < hitsNeeded && !js->isFuzzy)
{
warn("Error: %d of %d elements of %s.%s.%s are not in key %s.%s line %d of %s\n"
"Example miss: %s"
, total - hits, total, db, jf->table, jf->field
, keyField->table, keyField->field
, jf->lineIx, joiner->fileName, miss);
}
if (jf->unique || jf->full)
checkUniqueAndFull(joiner, js, db, jf, keyField, khiList);
}
freez(&miss);
slFreeList(&tableList);
sqlDisconnect(&conn);
}
}
