inc/wormdna.h File Reference

#include "dnautil.h"
#include "dnaseq.h"
#include "nt4.h"
#include "gdf.h"

Include dependency graph for wormdna.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Data Structures
struct	wormCdnaInfo
struct	wormCdnaIterator
struct	wormFeature
struct	wormGdfCache
Functions
boolean	wormCdnaInfo (char *name, struct wormCdnaInfo *retInfo)
void	wormFaCommentIntoInfo (char *faComment, struct wormCdnaInfo *retInfo)
void	wormFreeCdnaInfo (struct wormCdnaInfo *ci)
boolean	wormInfoForGene (char *orfName, struct wormCdnaInfo *retInfo)
boolean	wormCdnaSeq (char *name, struct dnaSeq **retDna, struct wormCdnaInfo *retInfo)
void	wormCdnaUncache ()
boolean	wormSearchAllCdna (struct wormCdnaIterator **retSi)
void	freeWormCdnaIterator (struct wormCdnaIterator **pIt)
dnaSeq *	nextWormCdna (struct wormCdnaIterator *it)
boolean	nextWormCdnaAndInfo (struct wormCdnaIterator *it, struct dnaSeq **retSeq, struct wormCdnaInfo *retInfo)
char *	wormFeaturesDir ()
char *	wormChromDir ()
char *	wormCdnaDir ()
char *	wormSangerDir ()
char *	wormGenieDir ()
char *	wormXenoDir ()
boolean	wormIsGeneName (char *name)
boolean	wormIsOrfName (char *in)
slName *	wormGeneToOrfNames (char *name)
char *	wormGeneFirstOrfName (char *geneName)
boolean	wormGeneForOrf (char *orfName, char *geneNameBuf, int bufSize)
boolean	getWormGeneExonDna (char *name, DNA **retDna)
boolean	getWormGeneDna (char *name, DNA **retDna, boolean upcExons)
void	wormLoadNt4Genome (struct nt4Seq ***retNt4Seq, int *retNt4Count)
void	wormFreeNt4Genome (struct nt4Seq ***pNt4Seq)
int	wormChromSize (char *chrom)
DNA *	wormChromPart (char *chromId, int start, int size)
DNA *	wormChromPartExonsUpper (char *chromId, int start, int size)
void	wormChromNames (char ***retNames, int *retNameCount)
int	wormChromIx (char *name)
char *	wormChromForIx (int ix)
char *	wormOfficialChromName (char *name)
boolean	wormGeneRange (char *name, char **retChromId, char *retStrand, int *retStart, int *retEnd)
boolean	wormParseChromRange (char *in, char **retChromId, int *retStart, int *retEnd)
boolean	wormIsChromRange (char *in)
void	wormClipRangeToChrom (char *chrom, int *pStart, int *pEnd)
boolean	wormIsNamelessCluster (char *name)
DNA *	wormGetNamelessClusterDna (char *name)
wormFeature *	newWormFeature (char *name, char *chrom, int start, int end, char typeByte)
wormFeature *	wormCdnasInRange (char *chromId, int start, int end)
wormFeature *	wormGenesInRange (char *chromId, int start, int end)
wormFeature *	wormSomeGenesInRange (char *chromId, int start, int end, char *gdfDir)
wormFeature *	wormCosmidsInRange (char *chromId, int start, int end)
cdaAli *	wormCdaAlisInRange (char *chromId, int start, int end)
FILE *	wormOpenGoodAli ()
gdfGene *	wormGetGdfGene (char *name)
gdfGene *	wormGetSomeGdfGene (char *name, struct wormGdfCache *cache)
gdfGene *	wormGetGdfGeneList (char *baseName, int baseNameSize)
gdfGene *	wormGetSomeGdfGeneList (char *baseName, int baseNameSize, struct wormGdfCache *cache)
gdfGene *	wormGdfGenesInRange (char *chrom, int start, int end, struct wormGdfCache *geneFinder)
void	wormUncacheGdf ()
void	wormUncacheSomeGdf (struct wormGdfCache *cache)
Variables
wormGdfCache	wormSangerGdfCache
wormGdfCache	wormGenieGdfCache

---

Function Documentation

void freeWormCdnaIterator

(

struct wormCdnaIterator **

pIt

)

Definition at line 478 of file wormdna.c.

References carefulClose(), wormCdnaIterator::faFile, and freez().

{
struct wormCdnaIterator *it = *pIt;
if (it != NULL)
    {
    carefulClose(&it->faFile);
    freez(pIt);
    }
}

Here is the call graph for this function:

boolean getWormGeneDna	(	char *	name,
		DNA **	retDna,
		boolean	upcExons
	)

Definition at line 656 of file wormdna.c.

References chromIds, gdfGene::chromIx, gdfGene::dataCount, gdfGene::dataPoints, FALSE, gdfFreeGene(), gdfGeneExtents(), gdfOffsetGene(), gdfRcGene(), slName::name, reverseComplement(), gdfGene::strand, toUpperN(), TRUE, wormChromPart(), wormGeneToOrfNames(), wormGenieGdfCache, wormGetSomeGdfGene(), wormIsGeneName(), and wormSangerGdfCache.

{
struct gdfGene *g;
struct slName *syn = NULL;
long lstart, lend;
int start, end;
int dnaSize;
DNA *dna;
struct wormGdfCache *gdfCache;

/* Translate biologist type name to cosmid.N name */
if (wormIsGeneName(name))
    {
    syn = wormGeneToOrfNames(name);
    if (syn != NULL)
        name = syn->name;
    }
if (strncmp(name, "g-", 2) == 0)
    gdfCache = &wormGenieGdfCache;
else
    gdfCache = &wormSangerGdfCache;
if ((g = wormGetSomeGdfGene(name, gdfCache)) == NULL)
    return FALSE;
gdfGeneExtents(g, &lstart, &lend);
start = lstart;
end = lend;
/* wormClipRangeToChrom(chromIds[g->chromIx], &start, &end); */
dnaSize = end-start;
*retDna = dna = wormChromPart(chromIds[g->chromIx], start, dnaSize);

gdfOffsetGene(g, -start);
if (g->strand == '-')
    {
    reverseComplement(dna, dnaSize);
    gdfRcGene(g, dnaSize);
    }
if (upcExons)
    {
    int i;
    struct gdfDataPoint *pt = g->dataPoints;
    for (i=0; i<g->dataCount; i += 2)
        {
        toUpperN(dna + pt[i].start, pt[i+1].start - pt[i].start);
        }
    }
gdfFreeGene(g);
return TRUE;
}

Here is the call graph for this function:

boolean getWormGeneExonDna	(	char *	name,
		DNA **	retDna
	)

Definition at line 706 of file wormdna.c.

References chromIds, gdfGene::chromIx, cloneString(), gdfGene::dataCount, gdfGene::dataPoints, dyStringAppendN(), dyStringFree, FALSE, gdfFreeGene(), gdfGeneExtents(), gdfOffsetGene(), gdfRcGene(), slName::name, newDyString(), reverseComplement(), gdfGene::strand, dyString::string, TRUE, wormChromPart(), wormGeneToOrfNames(), wormGenieGdfCache, wormGetSomeGdfGene(), wormIsGeneName(), and wormSangerGdfCache.

{
struct gdfGene *g;
struct slName *syn = NULL;
long lstart, lend;
int start, end;
int dnaSize;
DNA *dna;
int i;
struct gdfDataPoint *pt = NULL;
struct wormGdfCache *gdfCache;
struct dyString *dy = newDyString(1000);
/* Translate biologist type name to cosmid.N name */
if (wormIsGeneName(name))
    {
    syn = wormGeneToOrfNames(name);
    if (syn != NULL)
        name = syn->name;
    }
if (strncmp(name, "g-", 2) == 0)
    gdfCache = &wormGenieGdfCache;
else
    gdfCache = &wormSangerGdfCache;
if ((g = wormGetSomeGdfGene(name, gdfCache)) == NULL)
    return FALSE;
gdfGeneExtents(g, &lstart, &lend);
start = lstart;
end = lend;
/*wormClipRangeToChrom(chromIds[g->chromIx], &start, &end);*/
dnaSize = end-start;
dna = wormChromPart(chromIds[g->chromIx], start, dnaSize);

gdfOffsetGene(g, -start);
if (g->strand == '-')
    {
    reverseComplement(dna, dnaSize);
    gdfRcGene(g, dnaSize);
    }
pt = g->dataPoints;
for (i=0; i<g->dataCount; i += 2)
    {
    dyStringAppendN(dy, (dna+pt[i].start), (pt[i+1].start - pt[i].start));
    }
*retDna = cloneString(dy->string);
dyStringFree(&dy);
gdfFreeGene(g);
return TRUE;
}

Here is the call graph for this function:

struct wormFeature* newWormFeature	(	char *	name,
		char *	chrom,
		int	start,
		int	end,
		char	typeByte
	)			[read]

Definition at line 298 of file wormdna.c.

References wormFeature::chrom, wormFeature::end, name, wormFeature::name, needMem(), wormFeature::start, and wormFeature::typeByte.

Referenced by scanChromOffsetFile().

{
int size = sizeof(struct wormFeature) + strlen(name);
struct wormFeature *feat = needMem(size);
feat->chrom = chrom;
feat->start = start;
feat->end = end;
feat->typeByte = typeByte;
strcpy(feat->name, name);
return feat;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct dnaSeq* nextWormCdna

(

struct wormCdnaIterator *

it

)

[read]

Definition at line 458 of file wormdna.c.

References wormCdnaIterator::faFile, faReadOneDnaSeq(), and TRUE.

{
return faReadOneDnaSeq(it->faFile, "unknown", TRUE);
}

Here is the call graph for this function:

boolean nextWormCdnaAndInfo	(	struct wormCdnaIterator *	it,
		struct dnaSeq **	retSeq,
		struct wormCdnaInfo *	retInfo
	)

Definition at line 446 of file wormdna.c.

References wormCdnaIterator::faFile, FALSE, faReadNext(), TRUE, and wormFaCommentIntoInfo().

{
char *faComment;

if (!faReadNext(it->faFile, "unknown", TRUE, &faComment, retSeq))
    return FALSE;
wormFaCommentIntoInfo(faComment, retInfo);
return TRUE;
}

Here is the call graph for this function:

struct cdaAli* wormCdaAlisInRange	(	char *	chromId,
		int	start,
		int	end
	)			[read]

Definition at line 405 of file wormdna.c.

References AllocVar, alxSig, bits32, cdaLoadOne(), cdnaDir, errAbort(), mustOpen(), mustReadOne, readOne, SEEK_SET, slAddHead, slReverse(), and wormOpenGoodAli().

{
struct cdaAli *list = NULL, *el;
char fileName[512];
FILE *ixFile, *aliFile;
bits32 sig;
int s, e;
long fpos;

aliFile = wormOpenGoodAli();

sprintf(fileName, "%s%s.alx", cdnaDir, chromId);
ixFile = mustOpen(fileName, "rb");
mustReadOne(ixFile, sig);
if (sig != alxSig)
    errAbort("Bad signature on %s", fileName);

for (;;)
    {
    if (!readOne(ixFile, s))
        break;
    mustReadOne(ixFile, e);
    mustReadOne(ixFile, fpos);
    if (e <= start)
        continue;
    if (s >= end)
        break;
    AllocVar(el);
    fseek(aliFile, fpos, SEEK_SET);
    el = cdaLoadOne(aliFile);
    if (el == NULL)
        errAbort("Truncated cdnaAli file");
    slAddHead(&list, el);
    }
slReverse(&list);
fclose(aliFile);
fclose(ixFile);
return list;
}

Here is the call graph for this function:

char* wormCdnaDir

(

)

Definition at line 81 of file wormdna.c.

References cdnaDir, and getDirs().

{
getDirs();
return cdnaDir;
}

Here is the call graph for this function:

boolean wormCdnaInfo	(	char *	name,
		struct wormCdnaInfo *	retInfo
	)

Definition at line 261 of file wormdna.c.

References cdnaFa, cdnaSnof, cloneString(), errAbort(), FALSE, SEEK_SET, snofFindOffset(), TRUE, wormCdnaCache(), and wormFaCommentIntoInfo().

{
char commentBuf[512];
char *comment;
long offset;

wormCdnaCache();
if (!snofFindOffset(cdnaSnof, name, &offset))
    return FALSE;
fseek(cdnaFa, offset, SEEK_SET);
fgets(commentBuf, sizeof(commentBuf), cdnaFa);
if (commentBuf[0] != '>')
    errAbort("Expecting line starting with > in cDNA fa file.\nGot %s", commentBuf);
comment = cloneString(commentBuf);
wormFaCommentIntoInfo(comment, retInfo);
return TRUE;
}

Here is the call graph for this function:

boolean wormCdnaSeq	(	char *	name,
		struct dnaSeq **	retDna,
		struct wormCdnaInfo *	retInfo
	)

Definition at line 280 of file wormdna.c.

References cdnaFa, cdnaSnof, FALSE, faReadNext(), SEEK_SET, snofFindOffset(), TRUE, wormCdnaCache(), and wormFaCommentIntoInfo().

{
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;
}

Here is the call graph for this function:

struct wormFeature* wormCdnasInRange	(	char *	chromId,
		int	start,
		int	end
	)			[read]

Definition at line 358 of file wormdna.c.

References cdnaDir, cdoSig, getDirs(), and scanChromOffsetFile().

{
/* This routine looks through the .CDO files made by cdnaOff
 */
getDirs();
return scanChromOffsetFile(cdnaDir, ".cdo", cdoSig, 2*sizeof(int)+1, 
    chromId, start, end, 0);
}

Here is the call graph for this function:

void wormCdnaUncache

(

)

Definition at line 160 of file wormdna.c.

References carefulClose(), cdnaDir, cdnaFa, cdnaSnof, freez(), and snofClose().

{
snofClose(&cdnaSnof);
carefulClose(&cdnaFa);
freez(&cdnaDir);
}

Here is the call graph for this function:

char* wormChromDir

(

)

Definition at line 74 of file wormdna.c.

References getDirs(), and nt4Dir.

{
getDirs();
return nt4Dir;
}

Here is the call graph for this function:

char* wormChromForIx

(

int

ix

)

Definition at line 124 of file wormdna.c.

References ArraySize, and chromIds.

{
assert(ix >= 0 && ix <= ArraySize(chromIds));
return chromIds[ix];
}

int wormChromIx

(

char *

name

)

Definition at line 118 of file wormdna.c.

References chromIds, and stringIx.

Referenced by wormChromSize(), and wormOfficialChromName().

{
return stringIx(name, chromIds);
}

Here is the caller graph for this function:

void wormChromNames	(	char ***	retNames,
		int *	retNameCount
	)

Definition at line 111 of file wormdna.c.

References ArraySize, and chromIds.

Referenced by xenAlignWorm().

{
*retNames = chromIds;
*retNameCount = ArraySize(chromIds);
}

Here is the caller graph for this function:

DNA* wormChromPart	(	char *	chromId,
		int	start,
		int	size
	)

Definition at line 813 of file wormdna.c.

References makeChromFileName(), and nt4LoadPart().

Referenced by getWormGeneDna(), getWormGeneExonDna(), wormChromPartExonsUpper(), wormGetNamelessClusterDna(), and xenAlignWorm().

{
char fileName[512];
makeChromFileName(chromId, fileName);
return nt4LoadPart(fileName, start, size);
}

Here is the call graph for this function:

Here is the caller graph for this function:

DNA* wormChromPartExonsUpper	(	char *	chromId,
		int	start,
		int	size
	)

Definition at line 821 of file wormdna.c.

References gdfFreeGene(), gdfUpcExons(), name, wormFeature::name, wormFeature::next, slFreeList(), wormFeature::start, wormChromPart(), wormGenesInRange(), wormGetGdfGene(), and wormIsNamelessCluster().

{
DNA *dna = wormChromPart(chromId, start, size);
struct wormFeature *geneFeat = wormGenesInRange(chromId, start, start+size);
struct wormFeature *feat;

for (feat = geneFeat; feat != NULL; feat = feat->next)
    {
    char *name = feat->name;
    if (!wormIsNamelessCluster(name))
        {
        struct gdfGene *gene = wormGetGdfGene(name);
        gdfUpcExons(gene, feat->start, dna, size, start);
        gdfFreeGene(gene);
        }
    }
slFreeList(&geneFeat);
return dna;
}

Here is the call graph for this function:

int wormChromSize

(

char *

chrom

)

Definition at line 791 of file wormdna.c.

References ArraySize, chromIds, errAbort(), makeChromFileName(), nt4BaseCount(), and wormChromIx().

Referenced by wormClipRangeToChrom().

{
static int sizes[ArraySize(chromIds)];
int ix;
int size;

if ((ix = wormChromIx(chrom)) < 0)
    errAbort("%s isn't a chromosome", chrom);
size = sizes[ix];

/* If we don't know it already have to get it from file. */
if (size == 0)
    {
    char fileName[512];
    makeChromFileName(chromIds[ix], fileName);
    size = sizes[ix] = nt4BaseCount(fileName);
    }
return size;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void wormClipRangeToChrom	(	char *	chrom,
		int *	pStart,
		int *	pEnd
	)

Definition at line 842 of file wormdna.c.

References wormChromSize().

Referenced by wormParseChromRange(), and xenAlignWorm().

{
int chromEnd = wormChromSize(chrom);
int temp;

/* Swap ends if reversed. */
if (*pStart > *pEnd)
    {
    temp = *pEnd;
    *pEnd = *pStart;
    *pStart = temp;
    }
/* Generally speaking try to slide the range covered by
 * start-end inside the chromosome rather than just
 * truncating an end. */
if (*pStart < 0)
    {
    *pEnd -= *pStart;
    *pStart = 0;
    }
if (*pEnd > chromEnd)
    {
    *pStart -= *pEnd - chromEnd;
    *pEnd = chromEnd;
    }
/* This handles case where the range is larger than the chromosome. */
if (*pStart < 0)
    *pStart = 0;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct wormFeature* wormCosmidsInRange	(	char *	chromId,
		int	start,
		int	end
	)			[read]

Definition at line 385 of file wormdna.c.

References featDir, getDirs(), pgoSig, and scanChromOffsetFile().

{
/* This routine looks through the .COO files made by makePgo
 */
getDirs();
return scanChromOffsetFile(featDir, ".coo", pgoSig, 2*sizeof(int)+1,
    chromId, start, end, 1);
}

Here is the call graph for this function:

void wormFaCommentIntoInfo	(	char *	faComment,
		struct wormCdnaInfo *	retInfo
	)

Definition at line 241 of file wormdna.c.

References errAbort(), parseRestOfCdnaInfo(), and zeroBytes().

Referenced by nextWormCdnaAndInfo(), wormCdnaInfo(), wormCdnaSeq(), and wormInfoForGene().

{
if (retInfo)
    {
    char *s;
    zeroBytes(retInfo, sizeof(*retInfo));
    /* Separate out first word and use it as name. */
    s = strchr(faComment, ' ');
    if (s == NULL)
        errAbort("Expecting lots of info, just got %s", faComment);
    *s++ = 0;
    retInfo->name = faComment+1;
    retInfo->motherString = faComment;

    parseRestOfCdnaInfo(s, retInfo);
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

char* wormFeaturesDir

(

)

Definition at line 67 of file wormdna.c.

References featDir, and getDirs().

Referenced by wormGeneForOrf(), wormGeneToOrfNames(), and wormInfoForGene().

{
getDirs();
return featDir;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void wormFreeCdnaInfo

(

struct wormCdnaInfo *

ci

)

Definition at line 168 of file wormdna.c.

References freeMem(), wormCdnaInfo::motherString, and zeroBytes().

{
freeMem(ci->motherString);
zeroBytes(ci, sizeof(*ci));
}

Here is the call graph for this function:

void wormFreeNt4Genome

(

struct nt4Seq ***

pNt4Seq

)

Definition at line 779 of file wormdna.c.

References ArraySize, chromIds, freeNt4(), and freez().

Referenced by xenAlignWorm().

{
struct nt4Seq **seqs;
int i;
if ((seqs = *pNt4Seq) == NULL)
    return;
for (i=0; i<ArraySize(chromIds); ++i)
    freeNt4(&seqs[i]);
freez(pNt4Seq);
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct gdfGene* wormGdfGenesInRange	(	char *	chrom,
		int	start,
		int	end,
		struct wormGdfCache *	geneFinder
	)			[read]

Definition at line 1183 of file wormdna.c.

References errAbort(), name, slAddHead, slFreeList(), slReverse(), wormGenieDir(), wormGenieGdfCache, wormGetSomeGdfGene(), wormIsNamelessCluster(), wormSangerDir(), wormSangerGdfCache, and wormSomeGenesInRange().

{
char *dir = NULL;
struct gdfGene *gdfList = NULL, *gdf;
struct wormFeature *nameList, *name;

if (geneFinder == &wormSangerGdfCache)
    dir = wormSangerDir();
else if (geneFinder == &wormGenieGdfCache)
    dir = wormGenieDir();
else
    errAbort("Unknown geneFinder line %d of %s", __LINE__, __FILE__);

nameList = wormSomeGenesInRange(chrom, start, end, dir);
for (name = nameList; name != NULL; name = name->next)
    {
    char *n = name->name;
    if (!wormIsNamelessCluster(n))
        {
        gdf = wormGetSomeGdfGene(n, geneFinder);
        slAddHead(&gdfList, gdf);
        }
    }
slFreeList(&nameList);
slReverse(&gdfList);
return gdfList;
}

Here is the call graph for this function:

char* wormGeneFirstOrfName

(

char *

geneName

)

Definition at line 578 of file wormdna.c.

References cloneString(), name, slName::name, slFreeList(), and wormGeneToOrfNames().

Referenced by wormInfoForGene().

{
struct slName *synList = wormGeneToOrfNames(geneName);
char *name;
if (synList == NULL)
    return NULL;
name = cloneString(synList->name);
slFreeList(&synList);
return name;
}

Here is the call graph for this function:

Here is the caller graph for this function:

boolean wormGeneForOrf	(	char *	orfName,
		char *	geneNameBuf,
		int	bufSize
	)

Definition at line 590 of file wormdna.c.

References chopLine, FALSE, mustOpen(), TRUE, and wormFeaturesDir().

{
FILE *f;
char fileName[512];
char lineBuf[512];
int nameLen = strlen(orfName);
boolean ok = FALSE;

sprintf(fileName, "%sorf2gene", wormFeaturesDir());
f = mustOpen(fileName, "r");
while (fgets(lineBuf, sizeof(lineBuf), f))
    {
    if (strncmp(lineBuf, orfName, nameLen) == 0 && lineBuf[nameLen] == ' ')
        {
        char *words[2];
        int wordCount;
        wordCount = chopLine(lineBuf, words);
        assert((int)strlen(words[1]) < bufSize);
        strncpy(geneNameBuf, words[1], bufSize);
        ok = TRUE;
        break;
        }
    }
fclose(f);
return ok;
}

Here is the call graph for this function:

boolean wormGeneRange	(	char *	name,
		char **	retChromId,
		char *	retStrand,
		int *	retStart,
		int *	retEnd
	)

Definition at line 991 of file wormdna.c.

References chopLine, FALSE, featDir, findAltSpliceRange(), getDirs(), mustOpen(), slName::name, SEEK_SET, slFreeList(), snofFindOffset(), snofMustOpen(), touppers(), wormFixupOrfName(), wormGeneToOrfNames(), wormIsChromRange(), wormIsGeneName(), and wormParseChromRange().

Referenced by wormGetNamelessClusterDna().

{
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);
    fgets(lineBuf, sizeof(lineBuf), c2cFile);
    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;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct wormFeature* wormGenesInRange	(	char *	chromId,
		int	start,
		int	end
	)			[read]

Definition at line 376 of file wormdna.c.

References getDirs(), sangerDir, and wormSomeGenesInRange().

Referenced by wormChromPartExonsUpper().

{
/* This routine looks through the .PGO files made by makePgo
 */
getDirs();
return wormSomeGenesInRange(chromId, start, end, sangerDir);
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct slName* wormGeneToOrfNames

(

char *

name

)

[read]

Definition at line 539 of file wormdna.c.

References ArraySize, chopString(), errAbort(), errno, newSlName(), slAddTail(), tolowers(), whiteSpaceChopper, and wormFeaturesDir().

Referenced by getWormGeneDna(), getWormGeneExonDna(), wormGeneFirstOrfName(), and wormGeneRange().

{
struct slName *synList = NULL;
char synFileName[512];
FILE *synFile;
char lineBuf[128];
int nameLen = strlen(name);

/* genes are supposed to be lower case. */
tolowers(name);

/* Open synonym file and loop through each line of it */
sprintf(synFileName, "%ssyn", wormFeaturesDir());
if ((synFile = fopen(synFileName, "r")) == NULL)
    errAbort("Can't find synonym file '%s'. (errno: %d)\n", synFileName, errno);
while (fgets(lineBuf, ArraySize(lineBuf), synFile))
    {
    /* If first part of line matches chop up line. */
    if (strncmp(name, lineBuf, nameLen) == 0)
        {
        char *syns[32];
        int count;
        count = chopString(lineBuf, whiteSpaceChopper, syns, ArraySize(syns));

        /* Looks like we got a synonym.  Add all the aliases. */
        if (strcmp(name, syns[0]) == 0)
            {
            int i;
            for (i=1; i<count; ++i)
                slAddTail(&synList, newSlName(syns[i]));
            break;
            }
        }
    }
fclose(synFile);
return synList;
}

Here is the call graph for this function:

Here is the caller graph for this function:

char* wormGenieDir

(

)

Definition at line 95 of file wormdna.c.

References genieDir, and getDirs().

Referenced by wormGdfGenesInRange().

{
getDirs();
return genieDir;
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct gdfGene* wormGetGdfGene

(

char *

name

)

[read]

Definition at line 1139 of file wormdna.c.

References defaultGdfCache, and wormGetSomeGdfGene().

Referenced by wormChromPartExonsUpper().

{
return wormGetSomeGdfGene(name, defaultGdfCache);
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct gdfGene* wormGetGdfGeneList	(	char *	baseName,
		int	baseNameSize
	)			[read]

Definition at line 1177 of file wormdna.c.

References defaultGdfCache, and wormGetSomeGdfGeneList().

{
return wormGetSomeGdfGeneList(baseName, baseNameSize, defaultGdfCache);
}

Here is the call graph for this function:

DNA* wormGetNamelessClusterDna

(

char *

name

)

Definition at line 1070 of file wormdna.c.

References errAbort(), wormChromPart(), and wormGeneRange().

{
char *chrom;
int start, end;
char strand;
if (!wormGeneRange(name, &chrom, &strand, &start, &end))
    errAbort("Can't find %s in database", name);
return wormChromPart(chrom, start, end-start);
}

Here is the call graph for this function:

struct gdfGene* wormGetSomeGdfGene	(	char *	name,
		struct wormGdfCache *	cache
	)			[read]

Definition at line 1127 of file wormdna.c.

References wormGdfCache::file, gdfReadOneGene(), SEEK_SET, wormGdfCache::snof, snofFindOffset(), and wormCacheSomeGdf().

Referenced by getWormGeneDna(), getWormGeneExonDna(), wormGdfGenesInRange(), and wormGetGdfGene().

{
long offset;

wormCacheSomeGdf(cache);
if (!snofFindOffset(cache->snof, name, &offset) )
    return NULL;
fseek(cache->file, offset, SEEK_SET);
return gdfReadOneGene(cache->file);
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct gdfGene* wormGetSomeGdfGeneList	(	char *	baseName,
		int	baseNameSize,
		struct wormGdfCache *	cache
	)			[read]

Definition at line 1145 of file wormdna.c.

References wormGdfCache::file, gdfReadOneGene(), SEEK_SET, slAddTail(), slReverse(), wormGdfCache::snof, snofElementCount(), snofFindFirstStartingWith(), snofNameOffsetAtIx(), and wormCacheSomeGdf().

Referenced by wormGetGdfGeneList().

{
int snIx;
int maxIx;
struct snof *snof;
FILE *f;
struct gdfGene *list = NULL, *el;

wormCacheSomeGdf(cache);
snof = cache->snof;
f = cache->file;
if (!snofFindFirstStartingWith(snof, baseName, baseNameSize, &snIx))
    return NULL;

maxIx = snofElementCount(snof);
for (;snIx < maxIx; ++snIx)
    {
    long offset;
    char *geneName;

    snofNameOffsetAtIx(snof, snIx, &geneName, &offset);
    if (strncmp(geneName, baseName, baseNameSize) != 0)
        break;
    fseek(f, offset, SEEK_SET);
    el = gdfReadOneGene(f);
    slAddTail(&list, el);
    }
slReverse(&list);
return list;
}

Here is the call graph for this function:

Here is the caller graph for this function:

boolean wormInfoForGene	(	char *	orfName,
		struct wormCdnaInfo *	retInfo
	)

Definition at line 618 of file wormdna.c.

References cloneString(), FALSE, freeMem(), mustOpen(), TRUE, wormFaCommentIntoInfo(), wormFeaturesDir(), wormGeneFirstOrfName(), and wormIsGeneName().

{
FILE *f;
char fileName[512];
char lineBuf[512];
int nameLen;
char *info = NULL;
char *synName = NULL;
int lineCount = 0;

/* Make this one work for orfs as well as gene names */
if (wormIsGeneName(orfName))
    {
    synName = wormGeneFirstOrfName(orfName);
    if (synName != NULL)
        orfName = synName;
    }
sprintf(fileName, "%sorfInfo", wormFeaturesDir());
nameLen = strlen(orfName);
f = mustOpen(fileName, "r");
while (fgets(lineBuf, sizeof(lineBuf), f))
    {
    ++lineCount;
    if (strncmp(lineBuf, orfName, nameLen) == 0 && lineBuf[nameLen] == ' ')
        {
        info = cloneString(lineBuf);
        break;
        }
    }
freeMem(synName);
fclose(f);
if (info == NULL)
    return FALSE;
wormFaCommentIntoInfo(info, retInfo);
return TRUE;;
}

Here is the call graph for this function:

boolean wormIsChromRange

(

char *

in

)

Definition at line 898 of file wormdna.c.

References wormFeature::end, wormFeature::start, and wormParseChromRange().

Referenced by wormGeneRange().

{
char *chromId;
int start, end;
boolean ok;
ok =  wormParseChromRange(in, &chromId, &start, &end);
return ok;
}

Here is the call graph for this function:

Here is the caller graph for this function:

boolean wormIsGeneName

(

char *

name

)

Definition at line 518 of file wormdna.c.

References chopString(), FALSE, isAllAlpha(), and isAllDigit().

Referenced by getWormGeneDna(), getWormGeneExonDna(), wormGeneRange(), and wormInfoForGene().

{
char buf[128];
int partCount;
strncpy(buf, name, sizeof(buf));
partCount = chopString(buf, "-", NULL, 0);
if (partCount == 2)
    {
    char *parts[2];
    chopString(buf, "-", parts, 2);
    return isAllAlpha(parts[0]) && isAllDigit(parts[1]);
    }
else
    {
    return FALSE;
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

boolean wormIsNamelessCluster

(

char *

name

)

Definition at line 1057 of file wormdna.c.

References FALSE, and TRUE.

Referenced by wormChromPartExonsUpper(), and wormGdfGenesInRange().

{
char *e = strrchr(name, '.');
if (e == NULL)
    return FALSE;
if (e[1] != 'N')
    return FALSE;
if (!isdigit(e[2]))
    return FALSE;
return TRUE;
}

Here is the caller graph for this function:

boolean wormIsOrfName

(

char *

in

)

Definition at line 511 of file wormdna.c.

{
return strchr(in, '.') != NULL;
}

void wormLoadNt4Genome	(	struct nt4Seq ***	retNt4Seq,
		int *	retNt4Count
	)

Definition at line 762 of file wormdna.c.

References ArraySize, chromIds, loadNt4(), makeChromFileName(), and needMem().

Referenced by xenAlignWorm().

{
int count = ArraySize(chromIds);
struct nt4Seq **nt4s = needMem(count*sizeof(*nt4s));
int i;
char fileName[512];

for (i=0; i<count; ++i)
    {
    makeChromFileName(chromIds[i], fileName);
    nt4s[i] = loadNt4(fileName, chromIds[i]);
    }
*retNt4Seq = nt4s;
*retNt4Count = count;
}

Here is the call graph for this function:

Here is the caller graph for this function:

char* wormOfficialChromName

(

char *

name

)

Definition at line 131 of file wormdna.c.

References chromIds, and wormChromIx().

Referenced by wormParseChromRange().

{
int ix = wormChromIx(name);
if (ix < 0) return NULL;
return chromIds[ix];
}

Here is the call graph for this function:

Here is the caller graph for this function:

FILE* wormOpenGoodAli

(

)

Definition at line 395 of file wormdna.c.

References cdaOpenVerify(), cdnaDir, and getDirs().

Referenced by wormCdaAlisInRange().

{
char fileName[512];
getDirs();
sprintf(fileName, "%sgood.ali", cdnaDir);
return cdaOpenVerify(fileName);
}

Here is the call graph for this function:

Here is the caller graph for this function:

boolean wormParseChromRange	(	char *	in,
		char **	retChromId,
		int *	retStart,
		int *	retEnd
	)

Definition at line 873 of file wormdna.c.

References ArraySize, chopString(), FALSE, TRUE, wormClipRangeToChrom(), and wormOfficialChromName().

Referenced by findAltSpliceRange(), wormGeneRange(), and wormIsChromRange().

{
char *words[5];
int wordCount;
char *chromId;
char buf[128];

strncpy(buf, in, sizeof(buf));
wordCount = chopString(buf, "- \t\r\n:", words, ArraySize(words));
if (wordCount != 3)
    return FALSE;
chromId = wormOfficialChromName(words[0]);
if (chromId == NULL)
    return FALSE;
if (!isdigit(words[1][0]) || !isdigit(words[2][0]))
    return FALSE;
*retChromId = chromId;
*retStart = atoi(words[1]);
*retEnd = atoi(words[2]);
wormClipRangeToChrom(chromId, retStart, retEnd);
return TRUE;
}

Here is the call graph for this function:

Here is the caller graph for this function:

char* wormSangerDir

(

)

Definition at line 88 of file wormdna.c.

References getDirs(), and sangerDir.

Referenced by wormGdfGenesInRange().

{
getDirs();
return sangerDir;
}

Here is the call graph for this function:

Here is the caller graph for this function:

boolean wormSearchAllCdna

(

struct wormCdnaIterator **

retSi

)

Definition at line 464 of file wormdna.c.

References cdnaDir, getDirs(), mustOpen(), needMem(), and TRUE.

{
char buf[512];
struct wormCdnaIterator *it;

it = needMem(sizeof(*it));
getDirs();
sprintf(buf, "%s%s", cdnaDir, "allcdna.fa");
it->faFile = mustOpen(buf, "rb");
*retSi = it;
return TRUE;
}

Here is the call graph for this function:

struct wormFeature* wormSomeGenesInRange	(	char *	chromId,
		int	start,
		int	end,
		char *	gdfDir
	)			[read]

Definition at line 369 of file wormdna.c.

References pgoSig, and scanChromOffsetFile().

Referenced by wormGdfGenesInRange(), and wormGenesInRange().

{
return scanChromOffsetFile(gdfDir, ".pgo", pgoSig, 2*sizeof(int)+1,
    chromId, start, end, 0);
}

Here is the call graph for this function:

Here is the caller graph for this function:

void wormUncacheGdf

(

)

Definition at line 1121 of file wormdna.c.

References defaultGdfCache, and wormUncacheSomeGdf().

{
wormUncacheSomeGdf(defaultGdfCache);
}

Here is the call graph for this function:

void wormUncacheSomeGdf

(

struct wormGdfCache *

cache

)

Definition at line 1114 of file wormdna.c.

References carefulClose(), wormGdfCache::file, wormGdfCache::snof, and snofClose().

Referenced by wormUncacheGdf().

{
snofClose(&cache->snof);
carefulClose(&cache->file);
}

Here is the call graph for this function:

Here is the caller graph for this function:

char* wormXenoDir

(

)

Definition at line 102 of file wormdna.c.

References getDirs(), and xenoDir.

{
getDirs();
return xenoDir;
}

Here is the call graph for this function:

---

Variable Documentation

struct wormGdfCache wormGenieGdfCache

Definition at line 1082 of file wormdna.c.

Referenced by getWormGeneDna(), getWormGeneExonDna(), and wormGdfGenesInRange().

struct wormGdfCache wormSangerGdfCache

Definition at line 1081 of file wormdna.c.

Referenced by getWormGeneDna(), getWormGeneExonDna(), and wormGdfGenesInRange().

---