struct edwAssembly *edwAssemblyForUcscDb(struct sqlConnection *conn, char *ucscDb)
/* Get assembly for given UCSC ID or die trying */
{
char query[256];
sqlSafef(query, sizeof(query), "select * from edwAssembly where ucscDb='%s'", ucscDb);
struct edwAssembly *assembly = edwAssemblyLoadByQuery(conn, query);
if (assembly == NULL)
errAbort("Can't find assembly for %s", ucscDb);
return assembly;
}
void edwFixTargetSeq(char *when)
/* edwFixTargetSeq - Fill in new fields about target seq to edwBamFile and edwAssembly.. */
{
struct sqlConnection *conn = edwConnectReadWrite();
struct edwAssembly *as, *asList = edwAssemblyLoadByQuery(conn, "select * from edwAssembly");
char query[512];
for (as = asList; as != NULL; as = as->next)
{
char *twoBitFileName = edwPathForFileId(conn, as->twoBitId);
struct twoBitFile *tbf = twoBitOpen(twoBitFileName);
safef(query, sizeof(query), "update edwAssembly set seqCount=%u where id=%u",
tbf->seqCount, as->id);
sqlUpdate(conn, query);
freez(&twoBitFileName);
twoBitClose(&tbf);
}
edwAssemblyFreeList(&asList);
struct edwBamFile *bam, *bamList = edwBamFileLoadByQuery(conn, "select * from edwBamFile");
for (bam = bamList; bam != NULL; bam = bam->next)
{
char *fileName = edwPathForFileId(conn, bam->fileId);
samfile_t *sf = samopen(fileName, "rb", NULL);
if (sf == NULL)
errnoAbort("Couldn't open %s.\n", fileName);
bam_header_t *head = sf->header;
if (head == NULL)
errAbort("Aborting ... Bad BAM header in file: %s", fileName);
/* Sum up some target sizes. */
long long targetBaseCount = 0; /* Total size of all bases in target seq */
int i;
for (i=0; i<head->n_targets; ++i)
targetBaseCount += head->target_len[i];
safef(query, sizeof(query),
"update edwBamFile set targetBaseCount=%lld,targetSeqCount=%u where id=%u",
targetBaseCount, (unsigned)head->n_targets, bam->id);
sqlUpdate(conn, query);
samclose(sf);
freez(&fileName);
}
}
void screenFastqForContaminants(struct sqlConnection *conn,
struct edwFile *ef, struct edwValidFile *vf)
/* The ef/vf point to same file, which is fastq format. Set alignments up for a sample against all
* contamination targets. */
{
/* Get target list and see if we have any work to do. */
struct edwQaContamTarget *target, *targetList;
targetList = getContamTargets(conn, ef, vf);
boolean needScreen = FALSE;
for (target = targetList; target != NULL; target = target->next)
{
if (edwQaContamMade(conn, ef->id, target->id) <= 0)
{
needScreen = TRUE;
break;
}
}
if (needScreen)
{
verbose(1, "screenFastqForContaminants(%u(%s))\n", ef->id, ef->submitFileName);
/* Get fastq record. */
struct edwFastqFile *fqf = edwFastqFileFromFileId(conn, ef->id);
if (fqf == NULL)
errAbort("No edwFastqFile record for file id %lld", (long long)ef->id);
/* Create downsampled fastq in temp directory - downsampled more than default even. */
char sampleFastqName[PATH_LEN];
edwMakeTempFastqSample(fqf->sampleFileName, FASTQ_SAMPLE_SIZE, sampleFastqName);
verbose(1, "downsampled %s into %s\n", vf->licensePlate, sampleFastqName);
for (target = targetList; target != NULL; target = target->next)
{
/* Get assembly associated with target */
int assemblyId = target->assemblyId;
char query[512];
sqlSafef(query, sizeof(query), "select * from edwAssembly where id=%d", assemblyId);
struct edwAssembly *newAsm = edwAssemblyLoadByQuery(conn, query);
if (newAsm == NULL)
errAbort("warehouse edwQaContamTarget %d not found", assemblyId);
/* If we don't already have a match, do work to create contam record. */
int matchCount = edwQaContamMade(conn, ef->id, target->id);
if (matchCount <= 0)
{
/* We run the bed-file maker, just for side effect calcs. */
double mapRatio = 0, depth = 0, sampleCoverage = 0, uniqueMapRatio;
edwAlignFastqMakeBed(ef, newAsm, sampleFastqName, vf, NULL,
&mapRatio, &depth, &sampleCoverage, &uniqueMapRatio);
verbose(1, "%s mapRatio %g, depth %g, sampleCoverage %g\n",
newAsm->name, mapRatio, depth, sampleCoverage);
struct edwQaContam contam =
{.fileId=ef->id, .qaContamTargetId=target->id, .mapRatio = mapRatio};
edwQaContamSaveToDb(conn, &contam, "edwQaContam", 256);
}
edwAssemblyFree(&newAsm);
}
edwQaContamTargetFreeList(&targetList);
if (keepTemp)
verbose(1, "%s\n", sampleFastqName);
else
remove(sampleFastqName);
edwFastqFileFree(&fqf);
}
}