void cdwGroupUser(char *groupName, int userCount, char *userEmails[])
/* cdwGroupUser - Change user group settings.. */
{
struct sqlConnection *conn = cdwConnectReadWrite();
struct cdwGroup *group = cdwNeedGroupFromName(conn, groupName);
/* Build up array of all users, in the process aborting if user not
* found */
struct cdwUser *users[userCount];
int i;
for (i=0; i<userCount; ++i)
users[i] = cdwMustGetUserFromEmail(conn, userEmails[i]);
/* Now go through user by user adding things */
for (i=0; i<userCount; ++i)
{
struct cdwUser *user = users[i];
char query[256];
boolean inGroup = cdwUserInGroup(conn, user->id, group->id);
if (!inGroup && !clRemove)
{
sqlSafef(query, sizeof(query), "insert into cdwGroupUser (userId,groupId) values (%u,%u)",
user->id, group->id);
sqlUpdate(conn, query);
}
else if (inGroup && clRemove)
{
sqlSafef(query, sizeof(query), "delete from cdwGroupUser where userId=%u and groupId=%u",
user->id, group->id);
sqlUpdate(conn, query);
}
/* Deal with primary group */
if (clRemove)
{
if (group->id == user->primaryGroup)
{
/* If possible revert to another group. Otherwise will end up primary group 0 which
* is ok too. */
sqlSafef(query, sizeof(query),
"select groupId from cdwGroupUser where userId=%u and groupId != %u",
user->id, group->id);
int newPrimary = sqlQuickNum(conn, query);
sqlSafef(query, sizeof(query), "update cdwUser set primaryGroup=%d where id=%u",
newPrimary, user->id);
sqlUpdate(conn, query);
}
}
else
{
if (clPrimary || user->primaryGroup == 0)
{
sqlSafef(query, sizeof(query), "update cdwUser set primaryGroup=%u where id=%u",
group->id, user->id);
sqlUpdate(conn, query);
}
}
}
}
void cdwGroupFile(char *groupName, char *where)
/* cdwGroupFile - Associate a file with a group.. */
{
/* Get group from database, error out if no good */
struct sqlConnection *conn = cdwConnectReadWrite();
struct cdwGroup *group = cdwNeedGroupFromName(conn, groupName);
/* Get list of all stanzas matching query */
struct tagStorm *tags = cdwTagStorm(conn);
struct dyString *rqlQuery = dyStringNew(0);
dyStringPrintf(rqlQuery, "select accession from cdwFileTags where accession");
if (where != NULL)
dyStringPrintf(rqlQuery, " and %s", where);
struct slRef *ref, *matchRefList = tagStanzasMatchingQuery(tags, rqlQuery->string);
/* Make one pass through mostly for early error reporting and building up
* hash of cdwValidFiles keyed by accession */
struct hash *validHash = hashNew(0);
for (ref = matchRefList; ref != NULL; ref = ref->next)
{
struct tagStanza *stanza = ref->val;
char *acc = tagFindVal(stanza, "accession");
if (acc != NULL)
{
struct cdwValidFile *vf = cdwValidFileFromLicensePlate(conn, acc);
if (vf == NULL)
errAbort("%s not found in cdwValidFile", acc);
hashAdd(validHash, acc, vf);
}
}
/* Second pass through matching list we call routine that actually adds
* the group/file relationship. */
for (ref = matchRefList; ref != NULL; ref = ref->next)
{
struct tagStanza *stanza = ref->val;
char *acc = tagFindVal(stanza, "accession");
if (acc != NULL)
{
struct cdwValidFile *vf = hashFindVal(validHash, acc);
if (vf != NULL)
{
addGroupToValidFile(conn, vf, group);
}
}
}
if (clDry)
verbose(1, "Would have %s", (clRemove ? "removed" : "added"));
else
verbose(1, "%s", (clRemove ? "Removed" : "Added"));
verbose(1, " group %s to %d files\n", group->name, validHash->elCount);
}
void cdwMakeRepeatQa(int startFileId, int endFileId)
/* cdwMakeRepeatQa - Figure out what proportion of things align to repeats.. */
{
struct sqlConnection *conn = cdwConnectReadWrite();
struct cdwFile *ef, *efList = cdwFileAllIntactBetween(conn, startFileId, endFileId);
for (ef = efList; ef != NULL; ef = ef->next)
{
struct cdwValidFile *vf = cdwValidFileFromFileId(conn, ef->id);
if (vf != NULL)
{
if (sameString(vf->format, "fastq"))
fastqRepeatQa(conn, ef, vf);
}
}
sqlDisconnect(&conn);
}
void cdwMakePairedEndQa(unsigned startId, unsigned endId)
/* cdwMakePairedEndQa - Do alignments of paired-end fastq files and calculate distrubution of
* insert size. */
{
struct sqlConnection *conn = cdwConnectReadWrite();
struct cdwFile *ef, *efList = cdwFileAllIntactBetween(conn, startId, endId);
for (ef = efList; ef != NULL; ef = ef->next)
{
struct cdwValidFile *vf = cdwValidFileFromFileId(conn, ef->id);
if (vf != NULL)
{
if (sameString(vf->format, "fastq") && !isEmpty(vf->pairedEnd))
pairedEndQa(conn, ef, vf);
}
}
sqlDisconnect(&conn);
}
void cdwChangeAccess(char *chmodString, char *rqlWhere)
/* cdwChangeAccess - Change access to files.. */
{
char cWhere, cDir, cAccess;
parseChmodString(chmodString, &cWhere, &cDir, &cAccess);
/* Get list of all stanzas matching query */
struct sqlConnection *conn = cdwConnectReadWrite();
struct tagStorm *tags = cdwTagStorm(conn);
struct dyString *rqlQuery = dyStringNew(0);
dyStringPrintf(rqlQuery, "select accession from cdwFileTags where accession and %s", rqlWhere);
struct slRef *ref, *matchRefList = tagStanzasMatchingQuery(tags, rqlQuery->string);
/* Make one pass through mostly for early error reporting and building up
* hash of cdwValidFiles keyed by accession */
struct hash *validHash = hashNew(0);
for (ref = matchRefList; ref != NULL; ref = ref->next)
{
struct tagStanza *stanza = ref->val;
char *acc = tagFindVal(stanza, "accession");
if (acc != NULL)
{
struct cdwValidFile *vf = cdwValidFileFromLicensePlate(conn, acc);
if (vf == NULL)
errAbort("%s not found in cdwValidFile", acc);
hashAdd(validHash, acc, vf);
}
}
/* Second pass through matching list we call routine that actually adds
* the group/file relationship. */
for (ref = matchRefList; ref != NULL; ref = ref->next)
{
struct tagStanza *stanza = ref->val;
char *acc = tagFindVal(stanza, "accession");
if (acc != NULL)
{
struct cdwValidFile *vf = hashFindVal(validHash, acc);
if (vf != NULL)
{
changeAccess(conn, vf->fileId, cWhere, cDir, cAccess);
}
}
}
}
void cdwReallyRemoveFiles(char *email, char *submitUrl, int fileCount, char *fileIds[])
/* cdwReallyRemoveFiles - Remove files from data warehouse. Generally you want to depricate them
* instead. */
{
/* First convert all fileIds to binary. Do this first so bad command lines get caught. */
long long ids[fileCount];
int i;
for (i = 0; i<fileCount; ++i)
ids[i] = sqlLongLong(fileIds[i]);
/* Get hash of all submissions by user from that URL. Hash is keyed by ascii version of
* submitId. */
struct sqlConnection *conn = cdwConnectReadWrite();
struct cdwUser *user = cdwMustGetUserFromEmail(conn, email);
char query[256];
sqlSafef(query, sizeof(query),
" select cdwSubmit.id,cdwSubmitDir.id from cdwSubmit,cdwSubmitDir "
" where cdwSubmit.submitDirId=cdwSubmitDir.id and userId=%d "
" and cdwSubmitDir.url='%s' ",
user->id, submitUrl);
struct hash *submitHash = sqlQuickHash(conn, query);
/* Make sure that files and submission really go together. */
for (i=0; i<fileCount; ++i)
{
long long fileId = ids[i];
char buf[64];
sqlSafef(query, sizeof(query), "select submitId from cdwFile where id=%lld", fileId);
char *result = sqlQuickQuery(conn, query, buf, sizeof(buf));
if (result == NULL)
errAbort("%lld is not a fileId in the warehouse", fileId);
if (hashLookup(submitHash, result) == NULL)
errAbort("File ID %lld does not belong to submission set based on %s", fileId, submitUrl);
}
/* OK - paranoid checking is done, now let's remove each file from the tables it is in. */
for (i=0; i<fileCount; ++i)
{
cdwReallyRemoveFile(conn, ids[i], really);
}
}
void cdwUndeprecate(char *fileName)
/* cdwUndeprecate - Undeprecate a list of file accessions.. */
{
char **words, *buf;
int wordCount;
readAllWords(fileName, &words, &wordCount, &buf);
verbose(1, "Read %d accessions from %s\n", wordCount, fileName);
struct sqlConnection *conn = cdwConnectReadWrite();
int i;
for (i=0; i<wordCount; ++i)
{
char query[512];
sqlSafef(query, sizeof(query), "select * from cdwValidFile where licensePlate='%s'", words[i]);
struct cdwValidFile *vf = cdwValidFileLoadByQuery(conn, query);
if (vf == NULL)
errAbort("%s doesn't exist in cdwValidFile table", words[i]);
sqlSafef(query, sizeof(query), "update cdwFile set deprecated='' where id=%u", vf->fileId);
sqlUpdate(conn, query);
}
verbose(1, "Undeprecated %d files\n", i);
}
void pairedEndQa(struct sqlConnection *conn, struct cdwFile *ef, struct cdwValidFile *vf)
/* Look for other end, do a pairwise alignment, and save results in database. */
{
verbose(2, "pairedEndQa on %u %s %s\n", ef->id, ef->cdwFileName, ef->submitFileName);
/* Get other end, return if not found. */
struct cdwValidFile *otherVf = cdwOppositePairedEnd(conn, ef, vf);
if (otherVf == NULL)
return;
if (otherVf->fileId > vf->fileId)
return;
struct cdwValidFile *vf1, *vf2;
struct cdwQaPairedEndFastq *pair = cdwQaPairedEndFastqFromVfs(conn, vf, otherVf, &vf1, &vf2);
if (pair != NULL)
{
cdwValidFileFree(&otherVf);
return;
}
/* Get target assembly and figure out path for BWA index. */
struct cdwAssembly *assembly = cdwAssemblyForUcscDb(conn, vf->ucscDb);
assert(assembly != NULL);
char genoFile[PATH_LEN];
safef(genoFile, sizeof(genoFile), "%s%s/bwaData/%s.fa",
cdwValDataDir, assembly->ucscDb, assembly->ucscDb);
verbose(1, "aligning subsamples on %u vs. %u paired reads\n", vf1->fileId, vf2->fileId);
/* Make alignments of subsamples. */
char *sample1 = NULL, *sample2 = NULL, *sai1 = NULL, *sai2 = NULL;
makeTmpSai(conn, vf1, genoFile, &sample1, &sai1);
makeTmpSai(conn, vf2, genoFile, &sample2, &sai2);
/* Make paired end alignment */
char *tmpSam = cloneString(rTempName(cdwTempDir(), "cdwPairSample", ".sam"));
char command[6*PATH_LEN];
safef(command, sizeof(command),
"bwa sampe -n 1 -N 1 -f %s %s %s %s %s %s"
, tmpSam, genoFile, sai1, sai2, sample1, sample2);
mustSystem(command);
/* Make ra file with pairing statistics */
char *tmpRa = cloneString(rTempName(cdwTempDir(), "cdwPairSample", ".ra"));
safef(command, sizeof(command),
"edwSamPairedEndStats -maxInsert=%d %s %s", maxInsert, tmpSam, tmpRa);
mustSystem(command);
/* Read RA file into variables. */
struct cdwQaPairedEndFastq *pe = cdwQaPairedEndFastqOneFromRa(tmpRa);
/* Update database with record. */
struct sqlConnection *freshConn = cdwConnectReadWrite();
char query[256];
sqlSafef(query, sizeof(query),
"insert into cdwQaPairedEndFastq "
"(fileId1,fileId2,concordance,distanceMean,distanceStd,distanceMin,distanceMax,recordComplete) "
" values (%u,%u,%g,%g,%g,%g,%g,1)"
, vf1->fileId, vf2->fileId, pe->concordance, pe->distanceMean
, pe->distanceStd, pe->distanceMin, pe->distanceMax);
sqlUpdate(conn, query);
sqlDisconnect(&freshConn);
/* Clean up and go home. */
cdwValidFileFree(&otherVf);
remove(sample1);
remove(sample2);
remove(sai1);
remove(sai2);
remove(tmpSam);
remove(tmpRa);
#ifdef SOON
#endif /* SOON */
freez(&sample1);
freez(&sample2);
freez(&sai1);
freez(&sai2);
freez(&tmpSam);
freez(&tmpRa);
cdwQaPairedEndFastqFree(&pe);
cdwValidFileFree(&otherVf);
}
