boolean anyCdnaSeq(char *name, struct dnaSeq **retDna, struct wormCdnaInfo *retInfo)
/* Get a single cDNA sequence. Optionally (if retInfo is non-null) get additional
* info about the sequence. */
{
static FILE *cdnaFa;
static struct snof *cdnaSnof = NULL;
long offset;
char *faComment;
char **pFaComment = (retInfo == NULL ? NULL : &faComment);
if (cdnaSnof == NULL)
{
char buf[512];
cdnaSnof = snofMustOpen(cdnaName);
sprintf(buf, "%s%s", cdnaName, ".fa");
cdnaFa = mustOpen(buf, "rb");
}
if (!snofFindOffset(cdnaSnof, name, &offset))
return FALSE;
fseek(cdnaFa, offset, SEEK_SET);
if (!faReadNext(cdnaFa, name, TRUE, pFaComment, retDna))
return FALSE;
if (retInfo != NULL)
{
/* Kludge - only look up info if format is more or less right. */
int fieldCount = countChars(faComment, '|');
if (fieldCount >= 8)
wormFaCommentIntoInfo(faComment, retInfo);
else
zeroBytes(retInfo, sizeof(*retInfo));
}
return TRUE;
}
struct xaAli *getOneXaAli(char *organism, char *xaName)
/* Return a single named xaAli for organism. */
{
char ixFileName[512];
char dataFileName[512];
char *xDir;
struct snof *snof;
long offset;
FILE *f;
struct xaAli *xa;
xDir = wormXenoDir();
sprintf(ixFileName, "%s%s/all", xDir, organism);
sprintf(dataFileName, "%s%s/all%s", xDir, organism, xaAlignSuffix());
snof = snofMustOpen(ixFileName);
if (!snofFindOffset(snof, xaName, &offset))
errAbort("Couldn't find %s", xaName);
snofClose(&snof);
f = xaOpenVerify(dataFileName);
fseek(f, offset, SEEK_SET);
xa = xaReadNext(f, FALSE);
fclose(f);
return xa;
}
struct cdaAli *wormCdaAli(char *name)
/* Return named cDNA alignment or NULL if it doesn't exist. */
{
long offset;
wormOpenAliSnof();
if (!snofFindOffset(aliSnof, name, &offset))
return NULL;
fseek(aliFile, offset, SEEK_SET);
return cdaLoadOne(aliFile);
}
struct gdfGene *wormGetSomeGdfGene(char *name, struct wormGdfCache *cache)
/* Get a single gdfGene of given name. */
{
long offset;
wormCacheSomeGdf(cache);
if (!snofFindOffset(cache->snof, name, &offset) )
return NULL;
fseek(cache->file, offset, SEEK_SET);
return gdfReadOneGene(cache->file);
}
boolean wormCdnaSeq(char *name, struct dnaSeq **retDna, struct wormCdnaInfo *retInfo)
/* Get a single worm cDNA sequence. Optionally (if retInfo is non-null) get additional
* info about the sequence. */
{
long offset;
char *faComment;
char **pFaComment = (retInfo == NULL ? NULL : &faComment);
wormCdnaCache();
if (!snofFindOffset(cdnaSnof, name, &offset))
return FALSE;
fseek(cdnaFa, offset, SEEK_SET);
if (!faReadNext(cdnaFa, name, TRUE, pFaComment, retDna))
return FALSE;
wormFaCommentIntoInfo(faComment, retInfo);
return TRUE;
}
boolean wormCdnaInfo(char *name, struct wormCdnaInfo *retInfo)
/* Get info about cDNA sequence. */
{
char commentBuf[512];
char *comment;
long offset;
wormCdnaCache();
if (!snofFindOffset(cdnaSnof, name, &offset))
return FALSE;
fseek(cdnaFa, offset, SEEK_SET);
mustGetLine(cdnaFa, commentBuf, sizeof(commentBuf));
if (commentBuf[0] != '>')
errAbort("Expecting line starting with > in cDNA fa file.\nGot %s", commentBuf);
comment = cloneString(commentBuf);
wormFaCommentIntoInfo(comment, retInfo);
return TRUE;
}
boolean flyCdnaSeq(char *name, struct dnaSeq **retDna, struct wormCdnaInfo *retInfo)
/* Get a single fly cDNA sequence. Optionally (if retInfo is non-null) get additional
* info about the sequence. */
{
long offset;
char *faComment;
char **pFaComment = (retInfo == NULL ? NULL : &faComment);
static struct snof *cdnaSnof = NULL;
static FILE *cdnaFa;
if (cdnaSnof == NULL)
cdnaSnof = snofMustOpen("c:/biodata/fly/cDna/allcdna");
if (cdnaFa == NULL)
cdnaFa = mustOpen("c:/biodata/fly/cDna/allcdna.fa", "rb");
if (!snofFindOffset(cdnaSnof, name, &offset))
return FALSE;
fseek(cdnaFa, offset, SEEK_SET);
if (!faReadNext(cdnaFa, name, TRUE, pFaComment, retDna))
return FALSE;
flyFaCommentIntoInfo(faComment, retInfo);
return TRUE;
}
boolean wormGeneRange(char *name, char **retChrom, char *retStrand, int *retStart, int *retEnd)
/* Return chromosome position of a chrom range, gene, ORF, cosmid, or nameless cluster. */
{
static struct snof *c2gSnof = NULL, *c2cSnof = NULL;
static FILE *c2gFile = NULL, *c2cFile = NULL;
long offset;
char fileName[512];
struct slName *syn = NULL;
boolean ok;
if (wormIsChromRange(name))
{
*retStrand = '.';
return wormParseChromRange(name, retChrom, retStart, retEnd);
}
getDirs();
/* Translate biologist type name to cosmid.N name */
if (wormIsGeneName(name))
{
syn = wormGeneToOrfNames(name);
if (syn != NULL)
{
name = syn->name;
}
}
if (wormFixupOrfName(name)) /* See if ORF, and if so make nice. */
{
if (c2gSnof == NULL)
{
sprintf(fileName, "%sc2g", featDir);
c2gSnof = snofMustOpen(fileName);
sprintf(fileName, "%sc2g", featDir);
c2gFile = mustOpen(fileName, "rb");
}
ok = findAltSpliceRange(name, c2gSnof, c2gFile, retChrom, retStart, retEnd, retStrand);
}
else /* Lets say it's a cosmid. */
{
char lineBuf[128];
char *words[3];
int wordCount;
touppers(name);
if (c2cSnof == NULL)
{
sprintf(fileName, "%sc2c", featDir);
c2cSnof = snofMustOpen(fileName);
sprintf(fileName, "%sc2c", featDir);
c2cFile = mustOpen(fileName, "rb");
}
if (!snofFindOffset(c2cSnof, name, &offset) )
return FALSE;
fseek(c2cFile, offset, SEEK_SET);
mustGetLine(c2cFile, lineBuf, sizeof(lineBuf));
wordCount = chopLine(lineBuf, words);
assert(wordCount == 3);
assert(strcmp(words[2], name) == 0);
assert(wormIsChromRange(words[0]));
*retStrand = words[1][0];
ok = wormParseChromRange(words[0], retChrom, retStart, retEnd);
}
slFreeList(&syn);
return ok;
}
