static void getFastaOffsets(struct brokenRefPepTbl *brpTbl,
struct sqlConnection *conn,
struct extFileTbl* extFileTbl,
char *faPath)
/* parse fasta file to get offsets of proteins */
{
struct gbFa *fa = gbFaOpen(faPath, "r");
char acc[GB_ACC_BUFSZ];
struct brokenRefPep *brp;
HGID extId = extFileTblGet(extFileTbl, conn, faPath);
gbVerbMsg(5, "scanning fasta: %s", faPath);
while (gbFaReadNext(fa))
{
gbVerbMsg(5, " %s: %lld", fa->id, (long long)fa->recOff);
/* save only if same acecss, version, and file (to match mrna fa) */
short ver = gbSplitAccVer(fa->id, acc);
brp = hashFindVal(brpTbl->protAccHash, acc);
if ((brp != NULL) && (ver == brp->protVer) && sameString(faPath, brp->newFaPath))
{
gbFaGetSeq(fa); /* force read of sequence data */
brp->newFaId = extId;
brp->newFaOff = fa->recOff;
brp->newSeqSize = fa->seqLen;
brp->newRecSize = fa->off-fa->recOff;
gbVerbMsg(5, " save: %s %lld for %lld\n", fa->id, (long long)fa->recOff, (long long)fa->off);
}
}
gbFaClose(&fa);
}
void seqDataProcessUpdate(struct gbSelect* select)
/* Get sequences for a partition and update. Partition processed index should
* be loaded and selected versions flaged. */
{
char inFasta[PATH_LEN];
struct gbFa* inFa;
gbProcessedGetPath(select, "fa", inFasta);
inFa = gbFaOpen(inFasta, "r");
while (gbFaReadNext(inFa))
processSeq(select, inFa);
gbFaClose(&inFa);
}
boolean copyFastaRec(struct gbSelect* select, struct gbFa* inFa,
struct outFa* nativeFa, struct outFa* xenoFa)
/* Read and copy a record to one of the output files, if selected */
{
char acc[GB_ACC_BUFSZ];
unsigned version;
struct gbEntry* entry;
if (!gbFaReadNext(inFa))
return FALSE; /* EOF */
version = gbSplitAccVer(inFa->id, acc);
entry = gbReleaseFindEntry(select->release, acc);
if (entry != NULL)
{
char* seq = gbFaGetSeq(inFa);
if (strlen(seq) < MIN_SEQ_SIZE)
{
if (gbVerbose >= 3)
gbVerbPr(3, "skip %s, less than minimum sequence size", inFa->id);
}
else if ((version == entry->selectVer) && (entry->clientFlags & ALIGN_FLAG))
{
outFaWrite(((entry->orgCat == GB_NATIVE) ? nativeFa : xenoFa), inFa);
if (gbVerbose >= 3)
gbVerbPr(3, "aligning %s %s", inFa->id,
gbOrgCatName(entry->orgCat));
}
else if ((version == entry->selectVer) && (entry->clientFlags & MIGRATE_FLAG))
{
if (gbVerbose >= 3)
gbVerbPr(3, "migrating %s %s", inFa->id,
gbOrgCatName(entry->orgCat));
}
else
{
assert(version != entry->selectVer);
if (gbVerbose >= 3)
gbVerbPr(3, "skip %s, wrong version %s != %d",
gbOrgCatName(entry->orgCat), inFa->id,
entry->selectVer);
}
}
else
{
if (gbVerbose >= 3)
gbVerbPr(3, "skip %s, no entry", inFa->id);
}
return TRUE;
}
