static char *getCcdsRefSeqSummary(struct sqlConnection *conn, struct ccdsInfo *rsCcds)
/* get the refseq summary for a CCDS */
{
struct ccdsInfo *ci;
char accBuf[GENBANK_ACC_BUFSZ];
char *summary = NULL;
for (ci = rsCcds; ci != NULL; ci = ci->next)
{
summary = getRefSeqSummary(conn, genbankDropVer(accBuf, ci->mrnaAcc));
if (summary != NULL)
return summary;
}
return NULL;
}
static char *findRefSeqSummary(struct sqlConnection *conn,
struct geneSimilarities *refSeqs,
char **sumAccv)
/* Given similar refseq genes, find the first one with a RefSeq
* summary and return that summary, or NULL if not found. Also returns
* accv of matched */
{
char buf[GENBANK_ACC_BUFSZ];
struct geneSim *rs;
for (rs = refSeqs->genes; rs != NULL; rs = rs->next)
{
char *sum = getRefSeqSummary(conn, genbankDropVer(buf, rs->gene->name));
if (sum != NULL)
{
*sumAccv = cloneString(rs->gene->name);
return sum;
}
}
*sumAccv = NULL;
return NULL;
}
void hgKgGetText(char *database, char *outFile)
/* hgKgGetText - Get text from known genes into a file. */
{
FILE *f = mustOpen(outFile, "w");
struct sqlConnection *conn = sqlConnect(database);
struct sqlConnection *spConn = sqlConnect("uniProt");
struct sqlConnection *goConn = sqlConnect("go");
struct kgXref *kgList = NULL, *kg;
struct hash *refSeqHash = NULL;
/* Return hash keyed by refSeq NM_ id, with description values. */
gotRefSeqSummary = sqlTableExists(conn, summaryTable);
if (gotRefSeqSummary)
refSeqHash = getRefSeqSummary(conn);
else
warn("No %s table in %s, proceeding without...", summaryTable, database);
kgList = getKgList(conn);
verbose(1, "Read in %d known genes from %s\n", slCount(kgList), database);
for (kg = kgList; kg != NULL; kg = kg->next)
getText(kg, refSeqHash, conn, spConn, goConn, f);
carefulClose(&f);
}
