struct cdsPick *cdsPickLoadAllByChar(char *fileName, char chopper)
/* Load all cdsPick from a chopper separated file.
* Dispose of this with cdsPickFreeList(). */
{
struct cdsPick *list = NULL, *el;
struct lineFile *lf = lineFileOpen(fileName, TRUE);
char *row[12];
while (lineFileNextCharRow(lf, chopper, row, ArraySize(row)))
{
el = cdsPickLoad(row);
slAddHead(&list, el);
}
lineFileClose(&lf);
slReverse(&list);
return list;
}
struct cdsPick *cdsPickLoadAll(char *fileName)
/* Load all cdsPick from a whitespace-separated file.
* Dispose of this with cdsPickFreeList(). */
{
struct cdsPick *list = NULL, *el;
struct lineFile *lf = lineFileOpen(fileName, TRUE);
char *row[12];
while (lineFileRow(lf, row))
{
el = cdsPickLoad(row);
slAddHead(&list, el);
}
lineFileClose(&lf);
slReverse(&list);
return list;
}
void txGeneColor(char *uniProtDb, char *infoFile, char *pickFile, char *outFile)
/* txGeneColor - Figure out color to draw gene in.. */
{
/* Load picks into hash. We don't use cdsPicksLoadAll because empty fields
* cause that autoSql-generated routine problems. */
struct hash *pickHash = newHash(18);
struct cdsPick *pick;
struct lineFile *lf = lineFileOpen(pickFile, TRUE);
char *row[CDSPICK_NUM_COLS];
while (lineFileRowTab(lf, row))
{
pick = cdsPickLoad(row);
hashAdd(pickHash, pick->name, pick);
}
/* Open uniprot database connection. */
struct sqlConnection *uConn = sqlConnect(uniProtDb);
#ifdef OLD
/* Figure out our light and medium colors. */
mediumBlue.r = (6*trueBlue.r + 4*255)/10;
mediumBlue.g = (6*trueBlue.g + 4*255)/10;
mediumBlue.b = (6*trueBlue.b + 4*255)/10;
lightBlue.r = (1*trueBlue.r + 2*255)/3;
lightBlue.g = (1*trueBlue.g + 2*255)/3;
lightBlue.b = (1*trueBlue.b + 2*255)/3;
#endif /* OLD */
/* Read in info file, and loop through it to make out file. */
struct txInfo *info, *infoList = txInfoLoadAll(infoFile);
FILE *f = mustOpen(outFile, "w");
for (info = infoList; info != NULL; info = info->next)
{
struct rgbColor *col;
pick = hashFindVal(pickHash, info->name);
if (pick != NULL)
{
char *source = pick->source;
if (sameString(source, "RefPepValidated"))
col = &trueBlue;
else if (sameString(source, "ccds"))
col = &trueBlue;
else if (sameString(source, "RefPepReviewed"))
col = &trueBlue;
else if (sameString(source, "RefSeqValidated"))
col = &trueBlue;
else if (sameString(source, "RefSeqReviewed"))
col = &trueBlue;
else if (sameString(source, "swissProt"))
col = &trueBlue;
else if (startsWith("Ref", source))
col = &mediumBlue;
else
col = &lightBlue;
if (pick->swissProt[0] != 0)
{
char *acc = spLookupPrimaryAcc(uConn, pick->swissProt);
struct slName *pdbList = spPdbAccs(uConn, acc);
if (pdbList != NULL)
col = &black;
slFreeList(&pdbList);
}
}
else
col = &lightBlue;
fprintf(f, "%s\t%d\t%d\t%d\n", info->name, col->r, col->g, col->b);
}
carefulClose(&f);
}
void txGeneCdsMap(char *inBed, char *inInfo, char *inPicks, char *refPepToTxPsl,
char *refToPepTab, char *chromSizes, char *cdsToRna, char *rnaToGenome)
/* txGeneCdsMap - Create mapping between CDS region of gene and genome. */
{
/* Load info into hash. */
struct hash *infoHash = hashNew(18);
struct txInfo *info, *infoList = txInfoLoadAll(inInfo);
for (info = infoList; info != NULL; info = info->next)
hashAdd(infoHash, info->name, info);
/* Load picks into hash. We don't use cdsPicksLoadAll because empty fields
* cause that autoSql-generated routine problems. */
struct hash *pickHash = newHash(18);
struct cdsPick *pick;
struct lineFile *lf = lineFileOpen(inPicks, TRUE);
char *row[CDSPICK_NUM_COLS];
while (lineFileRowTab(lf, row))
{
pick = cdsPickLoad(row);
hashAdd(pickHash, pick->name, pick);
}
lineFileClose(&lf);
/* Load refPep/tx alignments into hash keyed by tx. */
struct hash *refPslHash = hashNew(18);
struct psl *psl, *pslList = pslLoadAll(refPepToTxPsl);
for (psl = pslList; psl != NULL; psl = psl->next)
hashAdd(refPslHash, psl->tName, psl);
struct hash *refToPepHash = hashTwoColumnFile(refToPepTab);
struct hash *chromSizeHash = hashNameIntFile(chromSizes);
/* Load in bed. */
struct bed *bed, *bedList = bedLoadNAll(inBed, 12);
/* Open output, and stream through bedList, writing output. */
FILE *fCdsToRna = mustOpen(cdsToRna, "w");
FILE *fRnaToGenome = mustOpen(rnaToGenome, "w");
int refTotal = 0, refFound = 0;
for (bed = bedList; bed != NULL; bed = bed->next)
{
if (bed->thickStart < bed->thickEnd)
{
char *chrom = bed->chrom;
int chromSize = hashIntVal(chromSizeHash, chrom);
info = hashMustFindVal(infoHash, bed->name);
pick = hashMustFindVal(pickHash, bed->name);
if (info->isRefSeq)
{
char *refAcc = txAccFromTempName(bed->name);
if (!startsWith("NM_", refAcc))
errAbort("Don't think I did find that refSeq acc, got %s", refAcc);
char *protAcc = hashMustFindVal(refToPepHash, refAcc);
++refTotal;
if (findAndMapPsl(bed, protAcc, refPslHash, chromSize, fCdsToRna))
++refFound;
}
else
{
fakeCdsToMrna(bed, fCdsToRna);
}
fakeRnaToGenome(bed, chromSize, fRnaToGenome);
}
}
verbose(1, "Missed %d of %d refSeq protein mappings. A small number of RefSeqs just map\n"
"to genome in the UTR.\n", refTotal - refFound, refTotal);
carefulClose(&fCdsToRna);
carefulClose(&fRnaToGenome);
}
void txGeneXref(char *genomeDb, char *uniProtDb, char *genePredFile, char *infoFile, char *pickFile,
char *evFile, char *outFile)
/* txGeneXref - Make kgXref type table for genes.. */
{
/* Load picks into hash. We don't use cdsPicksLoadAll because empty fields
* cause that autoSql-generated routine problems. */
struct hash *pickHash = newHash(18);
struct hash *geneToProtHash = makeGeneToProtHash(genePredFile);
struct cdsPick *pick;
struct lineFile *lf = lineFileOpen(pickFile, TRUE);
char *row[CDSPICK_NUM_COLS];
while (lineFileRowTab(lf, row))
{
pick = cdsPickLoad(row);
removePickVersions(pick);
hashAdd(pickHash, pick->name, pick);
}
/* Load evidence into hash */
struct hash *evHash = newHash(18);
struct txRnaAccs *ev, *evList = txRnaAccsLoadAll(evFile);
for (ev = evList; ev != NULL; ev = ev->next)
hashAdd(evHash, ev->name, ev);
/* Open connections to our databases */
struct sqlConnection *gConn = sqlConnect(genomeDb);
struct sqlConnection *uConn = sqlConnect(uniProtDb);
/* Read in info file, and loop through it to make out file. */
struct txInfo *info, *infoList = txInfoLoadAll(infoFile);
FILE *f = mustOpen(outFile, "w");
for (info = infoList; info != NULL; info = info->next)
{
char *kgID = info->name;
char *mRNA = "";
char *spID = "";
char *spDisplayID = "";
char *geneSymbol = NULL;
char *refseq = "";
char *protAcc = "";
char *description = NULL;
char query[256];
char *proteinId = hashMustFindVal(geneToProtHash, info->name);
boolean isAb = sameString(info->category, "antibodyParts");
pick = hashFindVal(pickHash, info->name);
ev = hashFindVal(evHash, info->name);
if (pick != NULL)
{
/* Fill in the relatively straightforward fields. */
refseq = pick->refSeq;
if (info->orfSize > 0)
{
protAcc = pick->refProt;
spID = proteinId;
if (sameString(protAcc, spID))
spID = pick->uniProt;
if (spID[0] != 0)
spDisplayID = spAnyAccToId(uConn, spID);
}
/* Fill in gene symbol and description from refseq if possible. */
if (refseq[0] != 0)
{
struct sqlResult *sr;
safef(query, sizeof(query), "select name,product from refLink where mrnaAcc='%s'",
refseq);
sr = sqlGetResult(gConn, query);
char **row = sqlNextRow(sr);
if (row != NULL)
{
geneSymbol = cloneString(row[0]);
if (!sameWord("unknown protein", row[1]))
description = cloneString(row[1]);
}
sqlFreeResult(&sr);
}
/* If need be try uniProt for gene symbol and description. */
if (spID[0] != 0 && (geneSymbol == NULL || description == NULL))
{
char *acc = spLookupPrimaryAcc(uConn, spID);
if (description == NULL)
description = spDescription(uConn, acc);
if (geneSymbol == NULL)
{
struct slName *nameList = spGenes(uConn, acc);
if (nameList != NULL)
geneSymbol = cloneString(nameList->name);
slFreeList(&nameList);
}
}
}
/* If it's an antibody fragment use that as name. */
if (isAb)
{
geneSymbol = cloneString("abParts");
description = cloneString("Parts of antibodies, mostly variable regions.");
isAb = TRUE;
}
if (ev == NULL)
{
mRNA = cloneString("");
if (!isAb)
{
errAbort("%s is %s but not %s\n", info->name, infoFile, evFile);
}
}
else
{
mRNA = cloneString(ev->primary);
chopSuffix(mRNA);
}
/* Still no joy? Try genbank RNA records. */
if (geneSymbol == NULL || description == NULL)
{
if (ev != NULL)
{
int i;
for (i=0; i<ev->accCount; ++i)
{
char *acc = ev->accs[i];
chopSuffix(acc);
if (geneSymbol == NULL)
{
safef(query, sizeof(query),
"select geneName.name from gbCdnaInfo,geneName "
"where geneName.id=gbCdnaInfo.geneName and gbCdnaInfo.acc = '%s'", acc);
geneSymbol = sqlQuickString(gConn, query);
if (geneSymbol != NULL)
{
if (sameString(geneSymbol, "n/a"))
geneSymbol = NULL;
}
}
if (description == NULL)
{
safef(query, sizeof(query),
"select description.name from gbCdnaInfo,description "
"where description.id=gbCdnaInfo.description "
"and gbCdnaInfo.acc = '%s'", acc);
description = sqlQuickString(gConn, query);
if (description != NULL)
{
if (sameString(description, "n/a"))
description = NULL;
}
}
}
}
}
if (geneSymbol == NULL)
geneSymbol = mRNA;
if (description == NULL)
description = mRNA;
/* Get rid of some characters that will cause havoc downstream. */
stripChar(geneSymbol, '\'');
subChar(geneSymbol, '<', '[');
subChar(geneSymbol, '>', ']');
/* Abbreviate geneSymbol if too long */
if (strlen(geneSymbol) > 40)
strcpy(geneSymbol+37, "...");
fprintf(f, "%s\t", kgID);
fprintf(f, "%s\t", mRNA);
fprintf(f, "%s\t", spID);
fprintf(f, "%s\t", spDisplayID);
fprintf(f, "%s\t", geneSymbol);
fprintf(f, "%s\t", refseq);
fprintf(f, "%s\t", protAcc);
fprintf(f, "%s\n", description);
}
carefulClose(&f);
}
void txGeneFromBed(char *inBed, char *inPicks, char *ucscFa, char *uniProtFa, char *refPepFa, char *outKg)
/* txGeneFromBed - Convert from bed to knownGenes format table (genePred + uniProt ID). */
{
/* Load protein sequence into hashes */
struct hash *uniProtHash = faReadAllIntoHash(uniProtFa, dnaUpper);
struct hash *ucscProtHash = faReadAllIntoHash(ucscFa, dnaUpper);
struct hash *refProtHash =faReadAllIntoHash(refPepFa, dnaUpper);
/* Load picks into hash. We don't use cdsPicksLoadAll because empty fields
* cause that autoSql-generated routine problems. */
struct hash *pickHash = newHash(18);
struct cdsPick *pick;
struct lineFile *lf = lineFileOpen(inPicks, TRUE);
char *row[CDSPICK_NUM_COLS];
while (lineFileRowTab(lf, row))
{
pick = cdsPickLoad(row);
hashAdd(pickHash, pick->name, pick);
}
/* Load in bed */
struct bed *bed, *bedList = bedLoadNAll(inBed, 12);
/* Do reformatting and write output. */
FILE *f = mustOpen(outKg, "w");
for (bed = bedList; bed != NULL; bed = bed->next)
{
char *protAcc = NULL;
if (bed->thickStart < bed->thickEnd)
{
pick = hashMustFindVal(pickHash, bed->name);
struct dnaSeq *spSeq = NULL, *uniSeq = NULL, *refPep = NULL, *ucscSeq;
ucscSeq = hashMustFindVal(ucscProtHash, bed->name);
if (pick->swissProt[0])
spSeq = hashMustFindVal(uniProtHash, pick->swissProt);
if (pick->uniProt[0])
uniSeq = hashMustFindVal(uniProtHash, pick->uniProt);
if (pick->refProt[0])
refPep = hashMustFindVal(refProtHash, pick->refProt);
/* First we look for an exact match between the ucsc protein and
* something from swissProt/uniProt. */
if (spSeq != NULL && sameString(ucscSeq->dna, spSeq->dna))
protAcc = pick->swissProt;
if (protAcc == NULL && uniSeq != NULL && sameString(ucscSeq->dna, uniSeq->dna))
protAcc = pick->uniProt;
if (protAcc == NULL && refPep != NULL && sameString(ucscSeq->dna, refPep->dna))
{
protAcc = cloneString(pick->refProt);
chopSuffix(protAcc);
}
if (protAcc == NULL)
{
if (pick->uniProt[0])
protAcc = pick->uniProt;
else
{
protAcc = cloneString(pick->refProt);
chopSuffix(protAcc);
}
}
}
outputKg(bed, emptyForNull(protAcc), f);
}
carefulClose(&f);
}
