static boolean isMgcFullLength()
/* determine if the current RA entry is for an MGC */
{
/* Check for keyword and source /clone field to identify MGC genes. Keywords
* end in `.' and are seperated by "; ". So look for .
* Clone can be like: /clone="MGC:9349 IMAGE:3846611" */
static boolean first = TRUE;
static regex_t keyRe, cloRe;
if (first)
{
reComp(&keyRe, "(^|.* )MGC(\\.$|;.*)");
reComp(&cloRe, "(^|.* )MGC:[0-9]+( .*|$)");
first = FALSE;
}
char *key = raFieldCurVal("key");
if (key == NULL)
return FALSE;
if (!reMatches(&keyRe, key))
return FALSE;
char *clo = raFieldCurVal("clo");
if (clo == NULL)
return FALSE;
if (!reMatches(&cloRe, clo))
return FALSE;
return TRUE;
}
static void imageCloneUpdate(struct gbStatus* status, struct sqlConnection *conn)
/* update image clone table */
{
/* assumes image id is never removed; most like true, would only
* be changed on a mistake */
if (status->stateChg & (GB_NEW|GB_META_CHG))
{
unsigned imageId = imageCloneGBParse(raFieldCurVal("clo"));
if (imageId != 0)
{
if (status->stateChg & GB_NEW)
{
imageCloneTblAdd(imageCloneTbl, imageId, status->acc, status->type,
raDir);
}
else if (status->stateChg & GB_META_CHG)
{
unsigned oldImageId = imageCloneTblGetId(conn, status->acc);
if (oldImageId == 0)
imageCloneTblAdd(imageCloneTbl, imageId, status->acc, status->type,
raDir);
else if (imageId != oldImageId)
imageCloneTblMod(imageCloneTbl, imageId, status->acc, raDir);
}
}
}
}
static boolean isMgcFullLength()
/* determine if the current RA entry is for an MGC */
{
/* n.b. this use to check for /clone= field, which was used when none full
* length MGCs were being processed. This is no longer needed now that the
* MGC project is over and we only check the keyword. */
return isMgcFullLengthKeyword(raFieldCurVal("key"));
}
static boolean isOrfeome()
/* determine if the current RA entry is for an ORFeome */
{
/* keyword containing "ORFeome collaboration" */
static boolean first = TRUE;
static regex_t keyRe;
if (first)
{
reComp(&keyRe, "(^|.* )ORFeome collaboration(\\.$|;.*)");
first = FALSE;
}
char *key = raFieldCurVal("key");
return (key != NULL) && reMatches(&keyRe, key);
}
static void refLinkUpdate(struct sqlConnection *conn, struct gbStatus* status)
/* Update the refLink table for the current entry */
{
int geneId;
char *gen = emptyForNull(raFieldCurVal("gen"));
char *pro = emptyForNull(raFieldCurVal("pro"));
gen = sqlEscapeString2(alloca(2*strlen(gen)+1), gen);
pro = sqlEscapeString2(alloca(2*strlen(pro)+1), pro);
/* can either have locus id (old locus link db) or gene id, or both,
* in which case the geneId is used */
geneId = (raGeneId != 0) ? raGeneId : raLocusLinkId;
if (status->stateChg & GB_NEW)
sqlUpdaterAddRow(refLinkUpd, "%s\t%s\t%s\t%s\t%u\t%u\t%u\t%u",
gen, pro, raAcc, raProtAcc, raFieldCurId("gen"),
raFieldCurId("pro"), geneId, raOmimId);
else if (status->stateChg & GB_META_CHG)
sqlUpdaterModRow(refLinkUpd, 1, "name='%s', product='%s', protAcc='%s', "
"geneName=%u, prodName=%u, locusLinkId=%u, "
"omimId=%u where mrnaAcc='%s'",
gen, pro, raProtAcc, raFieldCurId("gen"),
raFieldCurId("pro"), geneId, raOmimId, raAcc);
}
static void updateMetaData(struct sqlConnection *conn, struct gbStatus* status,
struct gbStatusTbl* statusTbl, HGID faFileId,
HGID pepFaId)
/* update the database tables for the current entry based on the stateChg
* flags */
{
assert(status->stateChg & (GB_NEW|GB_META_CHG|GB_REBUILD_DERIVED));
/* check for MGC, ORFeome */
if (status->orgCat == GB_NATIVE)
{
if (haveMgc)
status->isMgcFull = isMgcFullLength();
if (haveOrfeome)
status->isOrfeome = isOrfeome();
}
/* clear description if we are not keeping it */
if (!keepDesc(status))
raFieldClear("def");
/* most database changes are only done for GB_EXT_CHG */
if (status->stateChg & (GB_NEW|GB_META_CHG))
{
seqUpdate(status, faFileId); /* must be first to get status->gbSeqId */
gbCdnaInfoUpdate(status, conn);
imageCloneUpdate(status, conn);
}
if (raMiscDiffs != NULL)
gbMiscDiffUpdate(status, conn);
if (raWarn != NULL)
gbWarnUpdate(status, conn);
if ((gSrcDb == GB_REFSEQ) && (status->stateChg & (GB_NEW|GB_META_CHG)))
{
refSeqStatusUpdate(status);
refSeqSummaryUpdate(conn, status);
refLinkUpdate(conn, status);
refSeqPepUpdate(conn, pepFaId);
}
/* update in-memory gbStatus entry */
status->modDate = raModDate;
/* save CDS for use by the alignments */
if (!genbankCdsParse(raCds, &status->cds))
{
/* not valid CDS, only warn if RefSeq, where we expect to be better */
if ((gSrcDb == GB_REFSEQ) && gbIsProteinCodingRefSeq(status->acc))
prWarn("%s: malformed RefSeq CDS: %s", status->acc, raCds);
}
/* geneName for refFlat, if not available, try locus_tag */
char *geneName = raFieldCurVal("gen");
if (geneName == NULL)
geneName = raFieldCurVal("lot");
if (geneName != NULL)
status->geneName = lmCloneString(statusTbl->accHash->lm, geneName);
/* mark as done so dups in other updates don't get added to the database */
status->metaDone = TRUE;
}
static boolean isOrfeome()
/* determine if the current RA entry is for an ORFeome */
{
return isOrfeomeKeyword(raFieldCurVal("key"));
}
