void doBedReplicate(struct sqlConnection *conn, char *format, struct cdwAssembly *assembly,
struct cdwFile *elderEf, struct cdwValidFile *elderVf,
struct cdwFile *youngerEf, struct cdwValidFile *youngerVf)
/* Do correlation analysis between elder and younger and save result to
* a new cdwQaPairCorrelation record. Do this for a format where we have a bigBed file. */
{
/* If got both pairs, work is done already */
if (pairExists(conn, elderEf->id, youngerEf->id, "cdwQaPairSampleOverlap"))
return;
/* Get files for both younger and older. */
char *elderPath = cdwPathForFileId(conn, elderEf->id);
char *youngerPath = cdwPathForFileId(conn, youngerEf->id);
/* Do replicate calcs on bed3 lists from files. */
struct bed3 *elderBedList = bed3LoadAll(elderPath);
struct bed3 *youngerBedList = bed3LoadAll(youngerPath);
doBed3Replicate(conn, format, assembly, elderEf, elderVf, elderBedList,
youngerEf, youngerVf, youngerBedList);
/* Clean up. */
bed3FreeList(&elderBedList);
bed3FreeList(&youngerBedList);
freez(&youngerPath);
freez(&elderPath);
}
void doBigWigReplicate(struct sqlConnection *conn, struct edwAssembly *assembly,
struct edwFile *elderEf, struct edwValidFile *elderVf,
struct edwFile *youngerEf, struct edwValidFile *youngerVf)
/* Do correlation analysis between elder and younger and save result to
* a new edwQaPairCorrelation record. Do this for a format where we have a bigWig file. */
{
if (pairExists(conn, elderEf->id, youngerEf->id, "edwQaPairCorrelation"))
return;
char *enrichedIn = elderVf->enrichedIn;
if (!isEmpty(enrichedIn) && !sameString(enrichedIn, "unknown"))
{
struct genomeRangeTree *targetGrt = genomeRangeTreeForTarget(conn, assembly, enrichedIn);
/* Get open big wig files for both younger and older. */
char *elderPath = edwPathForFileId(conn, elderEf->id);
char *youngerPath = edwPathForFileId(conn, youngerEf->id);
struct bbiFile *elderBbi = bigWigFileOpen(elderPath);
struct bbiFile *youngerBbi = bigWigFileOpen(youngerPath);
/* Figure out thresholds */
double elderThreshold = twoStdsOverMean(elderBbi);
double youngerThreshold = twoStdsOverMean(youngerBbi);
/* Loop through a chromosome at a time adding to correlation, and at the end save result in r.*/
struct correlate *c = correlateNew(), *cInEnriched = correlateNew(), *cClipped = correlateNew();
struct bbiChromInfo *chrom, *chromList = bbiChromList(elderBbi);
struct bigWigValsOnChrom *aVals = bigWigValsOnChromNew();
struct bigWigValsOnChrom *bVals = bigWigValsOnChromNew();
for (chrom = chromList; chrom != NULL; chrom = chrom->next)
{
addBwCorrelations(chrom, targetGrt, aVals, bVals, elderBbi, youngerBbi,
elderThreshold, youngerThreshold, c, cInEnriched, cClipped);
}
/* Make up correlation structure . */
struct edwQaPairCorrelation *cor;
AllocVar(cor);
cor->elderFileId = elderVf->fileId;
cor->youngerFileId = youngerVf->fileId;
cor->pearsonOverall = correlateResult(c);
cor->pearsonInEnriched = correlateResult(cInEnriched);
cor->pearsonClipped = correlateResult(cClipped);
edwQaPairCorrelationSaveToDb(conn, cor, "edwQaPairCorrelation", 128);
bigWigValsOnChromFree(&bVals);
bigWigValsOnChromFree(&aVals);
genomeRangeTreeFree(&targetGrt);
freez(&cor);
correlateFree(&c);
bigWigFileClose(&youngerBbi);
bigWigFileClose(&elderBbi);
freez(&youngerPath);
freez(&elderPath);
}
}
void doSampleReplicate(struct sqlConnection *conn, char *format, struct cdwAssembly *assembly,
struct cdwFile *elderEf, struct cdwValidFile *elderVf,
struct cdwFile *youngerEf, struct cdwValidFile *youngerVf)
/* Do correlation analysis between elder and younger and save result to
* a new cdwQaPairSampleOverlap record. Do this for a format where we have a bed3 sample file. */
{
if (pairExists(conn, elderEf->id, youngerEf->id, "cdwQaPairSampleOverlap"))
return;
struct bed3 *elderBedList = cdwLoadSampleBed3(conn, elderVf);
struct bed3 *youngerBedList = cdwLoadSampleBed3(conn, youngerVf);
doBed3Replicate(conn, format, assembly, elderEf, elderVf, elderBedList,
youngerEf, youngerVf, youngerBedList);
bed3FreeList(&elderBedList);
bed3FreeList(&youngerBedList);
}
void doSampleReplicate(struct sqlConnection *conn, char *format, struct edwAssembly *assembly,
struct edwFile *elderEf, struct edwValidFile *elderVf,
struct edwFile *youngerEf, struct edwValidFile *youngerVf)
/* Do correlation analysis between elder and younger and save result to
* a new edwQaPairSampleOverlap record. Do this for a format where we have a bed3 sample file. */
{
if (pairExists(conn, elderEf->id, youngerEf->id, "edwQaPairSampleOverlap"))
return;
struct edwQaPairSampleOverlap *sam;
AllocVar(sam);
sam->elderFileId = elderVf->fileId;
sam->youngerFileId = youngerVf->fileId;
sam->elderSampleBases = elderVf->basesInSample;
sam->youngerSampleBases = youngerVf->basesInSample;
/* Load up elder into genome range tree. */
struct bed3 *elderBedList = edwLoadSampleBed3(conn, elderVf);
struct genomeRangeTree *elderGrt = edwMakeGrtFromBed3List(elderBedList);
/* Load up younger as bed, and loop through to get overlap */
long long totalOverlap = 0;
struct bed3 *bed, *youngerBedList = edwLoadSampleBed3(conn, youngerVf);
for (bed = youngerBedList; bed != NULL; bed = bed->next)
{
int overlap = genomeRangeTreeOverlapSize(elderGrt,
bed->chrom, bed->chromStart, bed->chromEnd);
totalOverlap += overlap;
}
sam->sampleOverlapBases = totalOverlap;
setSampleSampleEnrichment(sam, format, assembly, elderVf, youngerVf);
/* Save to database, clean up, go home. */
edwQaPairSampleOverlapSaveToDb(conn, sam, "edwQaPairSampleOverlap", 128);
freez(&sam);
genomeRangeTreeFree(&elderGrt);
bed3FreeList(&elderBedList);
bed3FreeList(&youngerBedList);
}
void doBigBedReplicate(struct sqlConnection *conn, char *format, struct edwAssembly *assembly,
struct edwFile *elderEf, struct edwValidFile *elderVf,
struct edwFile *youngerEf, struct edwValidFile *youngerVf)
/* Do correlation analysis between elder and younger and save result to
* a new edwQaPairCorrelation record. Do this for a format where we have a bigBed file. */
{
/* If got both pairs, work is done already */
if (pairExists(conn, elderEf->id, youngerEf->id, "edwQaPairSampleOverlap")
&& pairExists(conn, elderEf->id, youngerEf->id, "edwQaPairCorrelation"))
return;
int numColIx = 0;
if (sameString(format, "narrowPeak") || sameString(format, "broadPeak"))
numColIx = 6; // signalVal
else
numColIx = 4; // score
numColIx -= 3; // Subtract off chrom/start/end
char *enrichedIn = elderVf->enrichedIn;
struct genomeRangeTree *targetGrt = NULL;
if (!isEmpty(enrichedIn) && !sameString(enrichedIn, "unknown"))
targetGrt = genomeRangeTreeForTarget(conn, assembly, enrichedIn);
/* Get open big bed files for both younger and older. */
char *elderPath = edwPathForFileId(conn, elderEf->id);
char *youngerPath = edwPathForFileId(conn, youngerEf->id);
struct bbiFile *elderBbi = bigBedFileOpen(elderPath);
struct bbiFile *youngerBbi = bigBedFileOpen(youngerPath);
/* Loop through a chromosome at a time adding to correlation, and at the end save result in r.*/
struct correlate *c = correlateNew(), *cInEnriched = correlateNew();
struct bbiChromInfo *chrom, *chromList = bbiChromList(elderBbi);
long long elderTotalSpan = 0, youngerTotalSpan = 0, overlapTotalSpan = 0;
for (chrom = chromList; chrom != NULL; chrom = chrom->next)
{
addBbCorrelations(chrom, targetGrt, elderBbi, youngerBbi, numColIx, c, cInEnriched,
&elderTotalSpan, &youngerTotalSpan, &overlapTotalSpan);
}
/* Make up correlation structure and save. */
if (!pairExists(conn, elderEf->id, youngerEf->id, "edwQaPairCorrelation"))
{
struct edwQaPairCorrelation *cor;
AllocVar(cor);
cor->elderFileId = elderVf->fileId;
cor->youngerFileId = youngerVf->fileId;
cor->pearsonOverall = correlateResult(c);
cor->pearsonInEnriched = correlateResult(cInEnriched);
edwQaPairCorrelationSaveToDb(conn, cor, "edwQaPairCorrelation", 128);
freez(&cor);
}
/* Also make up sample structure and save. */
if (!pairExists(conn, elderEf->id, youngerEf->id, "edwQaPairSampleOverlap"))
{
struct edwQaPairSampleOverlap *sam;
AllocVar(sam);
sam->elderFileId = elderVf->fileId;
sam->youngerFileId = youngerVf->fileId;
sam->elderSampleBases = elderTotalSpan;
sam->youngerSampleBases = youngerTotalSpan;
sam->sampleOverlapBases = overlapTotalSpan;
setSampleSampleEnrichment(sam, format, assembly, elderVf, youngerVf);
edwQaPairSampleOverlapSaveToDb(conn, sam, "edwQaPairSampleOverlap", 128);
freez(&sam);
}
genomeRangeTreeFree(&targetGrt);
correlateFree(&c);
bigBedFileClose(&youngerBbi);
bigBedFileClose(&elderBbi);
freez(&youngerPath);
freez(&elderPath);
}
