lib/dnaMotif.c File Reference

#include "common.h"
#include "linefile.h"
#include "sqlList.h"
#include "dystring.h"
#include "dnaMotif.h"
#include "portable.h"
#include "dnaMotif.pss"

Include dependency graph for dnaMotif.c:

Go to the source code of this file.

Data Structures
struct	letterProb
Functions
dnaMotif *	dnaMotifCommaIn (char **pS, struct dnaMotif *ret)
void	dnaMotifFree (struct dnaMotif **pEl)
void	dnaMotifFreeList (struct dnaMotif **pList)
void	dnaMotifOutput (struct dnaMotif *el, FILE *f, char sep, char lastSep)
float	dnaMotifSequenceProb (struct dnaMotif *motif, DNA *dna)
char	dnaMotifBestStrand (struct dnaMotif *motif, DNA *dna)
double	dnaMotifBitScore (struct dnaMotif *motif, DNA *dna)
void	dnaMotifNormalize (struct dnaMotif *motif)
boolean	dnaMotifIsScoreBased (struct dnaMotif *motif)
void	dnaMotifScoreToProb (struct dnaMotif *motif)
void	dnaMotifMakeProbabalistic (struct dnaMotif *motif)
static void	printProbRow (FILE *f, char *label, float *p, int pCount)
void	dnaMotifPrintProb (struct dnaMotif *motif, FILE *f)
static double	u1 (double prob)
static double	uncertainty (struct dnaMotif *motif, int pos)
double	dnaMotifBitsOfInfo (struct dnaMotif *motif, int pos)
static struct letterProb *	letterProbNew (char letter, double prob)
static int	letterProbCmp (const void *va, const void *vb)
static void	addBaseProb (struct letterProb **pList, char letter, double prob)
static struct letterProb *	letterProbFromMotifColumn (struct dnaMotif *motif, int pos)
static void	psOneColumn (struct dnaMotif *motif, int pos, double xStart, double yStart, double width, double totalHeight, FILE *f)
static void	dnaMotifDims (struct dnaMotif *motif, double widthPerBase, double height, int *retWidth, int *retHeight)
void	dnaMotifToLogoPs (struct dnaMotif *motif, double widthPerBase, double height, char *fileName)
void	dnaMotifToLogoPng (struct dnaMotif *motif, double widthPerBase, double height, char *gsExe, char *tempDir, char *fileName)
void	dnaMotifToLogoPsW (struct dnaMotif *motif, double widthPerBase, double width, double height, char *fileName)
void	dnaMotifToLogoPGM (struct dnaMotif *motif, double widthPerBase, double width, double height, char *gsExe, char *tempDir, char *fileName)
Variables
static char const	rcsid [] = "$Id: dnaMotif.c,v 1.4 2006/09/14 21:08:24 braney Exp $"

---

Function Documentation

static void addBaseProb	(	struct letterProb **	pList,
		char	letter,
		double	prob
	)			[static]

Definition at line 333 of file dnaMotif.c.

References letterProbNew(), and slAddHead.

Referenced by letterProbFromMotifColumn().

{
if (prob > 0)
    {
    struct letterProb *lp = letterProbNew(letter, prob);
    slAddHead(pList, lp);
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

char dnaMotifBestStrand	(	struct dnaMotif *	motif,
		DNA *	dna
	)

Definition at line 173 of file dnaMotif.c.

References dnaMotif::columnCount, dnaMotifSequenceProb(), and reverseComplement().

{
float fScore, rScore;
fScore = dnaMotifSequenceProb(motif, dna);
reverseComplement(dna, motif->columnCount);
rScore = dnaMotifSequenceProb(motif, dna);
reverseComplement(dna, motif->columnCount);
if (fScore >= rScore)
    return '+';
else
    return '-';
}

Here is the call graph for this function:

double dnaMotifBitScore	(	struct dnaMotif *	motif,
		DNA *	dna
	)

Definition at line 187 of file dnaMotif.c.

References dnaMotif::columnCount, dnaMotifSequenceProb(), and logBase2.

{
double p = dnaMotifSequenceProb(motif, dna);
double q = pow(0.25, motif->columnCount);
double odds = p/q;
return logBase2(odds);
}

Here is the call graph for this function:

double dnaMotifBitsOfInfo	(	struct dnaMotif *	motif,
		int	pos
	)

Definition at line 293 of file dnaMotif.c.

References dnaMotif::columnCount, internalErr, and uncertainty().

Referenced by dnaMotifToLogoPs().

{
if (pos > motif->columnCount || pos < 0)
    internalErr();
return 2 - uncertainty(motif, pos);
}

Here is the call graph for this function:

Here is the caller graph for this function:

struct dnaMotif* dnaMotifCommaIn	(	char **	pS,
		struct dnaMotif *	ret
	)			[read]

Definition at line 14 of file dnaMotif.c.

References AllocArray, AllocVar, sqlEatChar(), sqlSignedComma(), and sqlStringComma().

{
char *s = *pS;
int i;

if (ret == NULL)
    AllocVar(ret);
ret->name = sqlStringComma(&s);
ret->columnCount = sqlSignedComma(&s);
s = sqlEatChar(s, '{');
AllocArray(ret->aProb, ret->columnCount);
for (i=0; i<ret->columnCount; ++i)
    {
    ret->aProb[i] = sqlSignedComma(&s);
    }
s = sqlEatChar(s, '}');
s = sqlEatChar(s, ',');
s = sqlEatChar(s, '{');
AllocArray(ret->cProb, ret->columnCount);
for (i=0; i<ret->columnCount; ++i)
    {
    ret->cProb[i] = sqlSignedComma(&s);
    }
s = sqlEatChar(s, '}');
s = sqlEatChar(s, ',');
s = sqlEatChar(s, '{');
AllocArray(ret->gProb, ret->columnCount);
for (i=0; i<ret->columnCount; ++i)
    {
    ret->gProb[i] = sqlSignedComma(&s);
    }
s = sqlEatChar(s, '}');
s = sqlEatChar(s, ',');
s = sqlEatChar(s, '{');
AllocArray(ret->tProb, ret->columnCount);
for (i=0; i<ret->columnCount; ++i)
    {
    ret->tProb[i] = sqlSignedComma(&s);
    }
s = sqlEatChar(s, '}');
s = sqlEatChar(s, ',');
*pS = s;
return ret;
}

Here is the call graph for this function:

static void dnaMotifDims	(	struct dnaMotif *	motif,
		double	widthPerBase,
		double	height,
		int *	retWidth,
		int *	retHeight
	)			[static]

Definition at line 380 of file dnaMotif.c.

References dnaMotif::columnCount.

Referenced by dnaMotifToLogoPGM(), dnaMotifToLogoPng(), dnaMotifToLogoPs(), and dnaMotifToLogoPsW().

{
static int widthFudgeFactor = 2, heightFudgeFactor = 2;
*retWidth = ceil(widthPerBase * motif->columnCount) + widthFudgeFactor;
*retHeight = ceil(height) + heightFudgeFactor;
}

Here is the caller graph for this function:

void dnaMotifFree

(

struct dnaMotif **

pEl

)

Definition at line 62 of file dnaMotif.c.

References dnaMotif::aProb, dnaMotif::cProb, freeMem(), freez(), dnaMotif::gProb, dnaMotif::name, and dnaMotif::tProb.

Referenced by dnaMotifFreeList().

{
struct dnaMotif *el;

if ((el = *pEl) == NULL) return;
freeMem(el->name);
freeMem(el->aProb);
freeMem(el->cProb);
freeMem(el->gProb);
freeMem(el->tProb);
freez(pEl);
}

Here is the call graph for this function:

Here is the caller graph for this function:

void dnaMotifFreeList

(

struct dnaMotif **

pList

)

Definition at line 77 of file dnaMotif.c.

References dnaMotifFree(), and dnaMotif::next.

{
struct dnaMotif *el, *next;

for (el = *pList; el != NULL; el = next)
    {
    next = el->next;
    dnaMotifFree(&el);
    }
*pList = NULL;
}

Here is the call graph for this function:

boolean dnaMotifIsScoreBased

(

struct dnaMotif *

motif

)

Definition at line 216 of file dnaMotif.c.

References dnaMotif::aProb, dnaMotif::columnCount, dnaMotif::cProb, FALSE, dnaMotif::gProb, dnaMotif::tProb, and TRUE.

Referenced by dnaMotifMakeProbabalistic().

{
int i;
for (i=0; i<motif->columnCount; ++i)
    {
    if (motif->aProb[i] < 0) return TRUE;
    if (motif->cProb[i] < 0) return TRUE;
    if (motif->gProb[i] < 0) return TRUE;
    if (motif->tProb[i] < 0) return TRUE;
    }
return FALSE;
}

Here is the caller graph for this function:

void dnaMotifMakeProbabalistic

(

struct dnaMotif *

motif

)

Definition at line 247 of file dnaMotif.c.

References dnaMotifIsScoreBased(), dnaMotifNormalize(), and dnaMotifScoreToProb().

{
if (dnaMotifIsScoreBased(motif))
    dnaMotifScoreToProb(motif);
else
    dnaMotifNormalize(motif);
}

Here is the call graph for this function:

void dnaMotifNormalize

(

struct dnaMotif *

motif

)

Definition at line 197 of file dnaMotif.c.

References dnaMotif::aProb, dnaMotif::columnCount, dnaMotif::cProb, errAbort(), dnaMotif::gProb, dnaMotif::name, and dnaMotif::tProb.

Referenced by dnaMotifMakeProbabalistic(), and dnaMotifScoreToProb().

{
int i;
for (i=0; i<motif->columnCount; ++i)
    {
    float sum = motif->aProb[i] + motif->cProb[i] + motif->gProb[i] + motif->tProb[i];
    if (sum < 0)
        errAbort("%s has negative numbers, perhaps it's score not probability based", 
                motif->name);
    if (sum == 0)
         motif->aProb[i] = motif->cProb[i] = motif->gProb[i] = motif->tProb[i] = 0.25;
    motif->aProb[i] /= sum;
    motif->cProb[i] /= sum;
    motif->gProb[i] /= sum;
    motif->tProb[i] /= sum;
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

void dnaMotifOutput	(	struct dnaMotif *	el,
		FILE *	f,
		char	sep,
		char	lastSep
	)

Definition at line 90 of file dnaMotif.c.

References dnaMotif::aProb, dnaMotif::columnCount, dnaMotif::cProb, dnaMotif::gProb, dnaMotif::name, and dnaMotif::tProb.

{
int i;
if (sep == ',') fputc('"',f);
fprintf(f, "%s", el->name);
if (sep == ',') fputc('"',f);
fputc(sep,f);
fprintf(f, "%d", el->columnCount);
fputc(sep,f);
if (sep == ',') fputc('{',f);
for (i=0; i<el->columnCount; ++i)
    {
    fprintf(f, "%f", el->aProb[i]);
    fputc(',', f);
    }
if (sep == ',') fputc('}',f);
fputc(sep,f);
if (sep == ',') fputc('{',f);
for (i=0; i<el->columnCount; ++i)
    {
    fprintf(f, "%f", el->cProb[i]);
    fputc(',', f);
    }
if (sep == ',') fputc('}',f);
fputc(sep,f);
if (sep == ',') fputc('{',f);
for (i=0; i<el->columnCount; ++i)
    {
    fprintf(f, "%f", el->gProb[i]);
    fputc(',', f);
    }
if (sep == ',') fputc('}',f);
fputc(sep,f);
if (sep == ',') fputc('{',f);
for (i=0; i<el->columnCount; ++i)
    {
    fprintf(f, "%f", el->tProb[i]);
    fputc(',', f);
    }
if (sep == ',') fputc('}',f);
fputc(lastSep,f);
}

void dnaMotifPrintProb	(	struct dnaMotif *	motif,
		FILE *	f
	)

Definition at line 267 of file dnaMotif.c.

References dnaMotif::aProb, dnaMotif::columnCount, dnaMotif::cProb, dnaMotif::gProb, printProbRow(), and dnaMotif::tProb.

{
printProbRow(f, "A", motif->aProb, motif->columnCount);
printProbRow(f, "C", motif->cProb, motif->columnCount);
printProbRow(f, "G", motif->gProb, motif->columnCount);
printProbRow(f, "T", motif->tProb, motif->columnCount);
}

Here is the call graph for this function:

void dnaMotifScoreToProb

(

struct dnaMotif *

motif

)

Definition at line 231 of file dnaMotif.c.

References dnaMotif::aProb, dnaMotif::columnCount, dnaMotif::cProb, dnaMotifNormalize(), dnaMotif::gProb, and dnaMotif::tProb.

Referenced by dnaMotifMakeProbabalistic().

                                    : 25% for each base */
{
int i;
for (i=0; i<motif->columnCount; ++i)
    {
    motif->aProb[i] = exp(motif->aProb[i]);
    motif->cProb[i] = exp(motif->cProb[i]);
    motif->gProb[i] = exp(motif->gProb[i]);
    motif->tProb[i] = exp(motif->tProb[i]);
    }
dnaMotifNormalize(motif);
}

Here is the call graph for this function:

Here is the caller graph for this function:

float dnaMotifSequenceProb	(	struct dnaMotif *	motif,
		DNA *	dna
	)

Definition at line 134 of file dnaMotif.c.

References dnaMotif::aProb, dnaMotif::columnCount, dnaMotif::cProb, dnaMotif::gProb, internalErr, dnaMotif::tProb, and warn().

Referenced by dnaMotifBestStrand(), and dnaMotifBitScore().

{
float p = 1.0;
int i;
for (i=0; i<motif->columnCount; ++i)
    {
    switch (dna[i])
        {
        case 'a':
        case 'A':
            p *= motif->aProb[i];
            break;
        case 'c':
        case 'C':
            p *= motif->cProb[i];
            break;
        case 'g':
        case 'G':
            p *= motif->gProb[i];
            break;
        case 't':
        case 'T':
            p *= motif->tProb[i];
            break;
        case 0:
            warn("dna shorter than motif");
            internalErr();
            break;
        default:
            p *= 0.25;
            break;
        }
    }
return p;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void dnaMotifToLogoPGM	(	struct dnaMotif *	motif,
		double	widthPerBase,
		double	width,
		double	height,
		char *	gsExe,
		char *	tempDir,
		char *	fileName
	)

Definition at line 484 of file dnaMotif.c.

References dnaMotifDims(), dnaMotifToLogoPsW(), dyStringAppend(), dyStringFree, dyStringNew, dyStringPrintf(), errAbort(), newDyString(), rTempName(), and dyString::string.

{
char *psName = rTempName(tempDir, "dnaMotif", ".ps");
struct dyString *dy = dyStringNew(0);
int w, h;
int sysRet;

if (gsExe == NULL) gsExe = "gs";
if (tempDir == NULL) tempDir = "/tmp";
dnaMotifToLogoPsW(motif, widthPerBase, width, height, psName);
dnaMotifDims(motif, widthPerBase, height, &w, &h);
dyStringAppend(dy, gsExe);
dyStringAppend(dy, " -sDEVICE=pgmraw -sOutputFile=");
dyStringAppend(dy, fileName);
dyStringAppend(dy, " -dBATCH -dNOPAUSE -q ");
dyStringPrintf(dy, "-g%dx%d ", (int) ceil(width), h);
dyStringAppend(dy, psName);
sysRet = system(dy->string);
if (sysRet != 0)
    errAbort("System call returned %d for:\n  %s", sysRet, dy->string);

/* Clean up. */
dyStringFree(&dy);

/* change permisssions so the webserver can access the file */
dy = newDyString(0);
dyStringPrintf(dy, "chmod 666 %s ", fileName);
sysRet = system(dy->string);

remove(psName);
}

Here is the call graph for this function:

void dnaMotifToLogoPng	(	struct dnaMotif *	motif,
		double	widthPerBase,
		double	height,
		char *	gsExe,
		char *	tempDir,
		char *	fileName
	)

Definition at line 417 of file dnaMotif.c.

References dnaMotifDims(), dnaMotifToLogoPs(), dyStringAppend(), dyStringFree, dyStringNew, dyStringPrintf(), errAbort(), newDyString(), rTempName(), and dyString::string.

{
char *psName = rTempName(tempDir, "dnaMotif", ".ps");
struct dyString *dy = dyStringNew(0);
int w, h;
int sysRet;

if (gsExe == NULL) gsExe = "gs";
if (tempDir == NULL) tempDir = "/tmp";
dnaMotifToLogoPs(motif, widthPerBase, height, psName);
dnaMotifDims(motif, widthPerBase, height, &w, &h);
dyStringAppend(dy, gsExe);
dyStringAppend(dy, " -sDEVICE=png16m -sOutputFile=");
dyStringAppend(dy, fileName);
dyStringAppend(dy, " -dBATCH -dNOPAUSE -q ");
dyStringPrintf(dy, "-g%dx%d ", w, h);
dyStringAppend(dy, psName);
sysRet = system(dy->string);
if (sysRet != 0)
    errAbort("System call returned %d for:\n  %s", sysRet, dy->string);

/* Clean up. */
dyStringFree(&dy);

/* change permisssions so the webserver can access the file */
dy = newDyString(0);
dyStringPrintf(dy, "chmod 666 %s ", fileName);
sysRet = system(dy->string);

remove(psName);
}

Here is the call graph for this function:

void dnaMotifToLogoPs	(	struct dnaMotif *	motif,
		double	widthPerBase,
		double	height,
		char *	fileName
	)

Definition at line 389 of file dnaMotif.c.

References carefulClose(), dnaMotif::columnCount, dnaMotifBitsOfInfo(), dnaMotifDims(), mustOpen(), and psOneColumn().

Referenced by dnaMotifToLogoPng().

{
FILE *f = mustOpen(fileName, "w");
int i;
int xStart = 0;
int w, h;
char *s = 
#include "dnaMotif.pss"
;

dnaMotifDims(motif, widthPerBase, height, &w, &h);
fprintf(f, "%%!PS-Adobe-3.1 EPSF-3.0\n");
fprintf(f, "%%%%BoundingBox: 0 0 %d %d\n\n", w, h);
fprintf(f, "%s", s);

fprintf(f, "%s", "% Start of code for this specific logo\n");

for (i=0; i<motif->columnCount; ++i)
    {
    double infoScale = dnaMotifBitsOfInfo(motif, i)/2.0;
    psOneColumn(motif, i, xStart, 0, widthPerBase, infoScale * height, f);
    xStart += widthPerBase;
    }
fprintf(f, "showpage\n");
carefulClose(&f);
}

Here is the call graph for this function:

Here is the caller graph for this function:

void dnaMotifToLogoPsW	(	struct dnaMotif *	motif,
		double	widthPerBase,
		double	width,
		double	height,
		char *	fileName
	)

Definition at line 456 of file dnaMotif.c.

References carefulClose(), dnaMotif::columnCount, dnaMotifDims(), mustOpen(), and psOneColumn().

Referenced by dnaMotifToLogoPGM().

{
FILE *f = mustOpen(fileName, "w");
int i;
int xStart = 0;
int w, h;
char *s = 
#include "dnaMotif.pss"
;

dnaMotifDims(motif, widthPerBase, height, &w, &h);
fprintf(f, "%%!PS-Adobe-3.1 EPSF-3.0\n");
fprintf(f, "%%%%BoundingBox: 0 0 %d %d\n\n", w, h);
fprintf(f, "%s", s);

fprintf(f, "%s", "% Start of code for this specific logo\n");

for (i=0; i<motif->columnCount; ++i)
    {
    double infoScale = 0.9 ; 
    xStart = i * width / motif->columnCount;
    psOneColumn(motif, i, xStart, 0, widthPerBase, infoScale * height, f);
    }
fprintf(f, "showpage\n");
carefulClose(&f);
}

Here is the call graph for this function:

Here is the caller graph for this function:

static int letterProbCmp	(	const void *	va,
		const void *	vb
	)			[static]

Definition at line 319 of file dnaMotif.c.

References letterProb::prob.

Referenced by letterProbFromMotifColumn().

{
const struct letterProb *a = *((struct letterProb **)va);
const struct letterProb *b = *((struct letterProb **)vb);
double dif = a->prob - b->prob;
if (dif < 0)
   return -1;
else if (dif > 0)
   return 1;
else
   return 0;
}

Here is the caller graph for this function:

static struct letterProb* letterProbFromMotifColumn	(	struct dnaMotif *	motif,
		int	pos
	)			[static, read]

Definition at line 343 of file dnaMotif.c.

References addBaseProb(), dnaMotif::aProb, dnaMotif::cProb, dnaMotif::gProb, letterProbCmp(), slSort(), and dnaMotif::tProb.

Referenced by psOneColumn().

{
struct letterProb *lpList = NULL;
addBaseProb(&lpList, 'A', motif->aProb[pos]);
addBaseProb(&lpList, 'C', motif->cProb[pos]);
addBaseProb(&lpList, 'G', motif->gProb[pos]);
addBaseProb(&lpList, 'T', motif->tProb[pos]);
slSort(&lpList, letterProbCmp);
return lpList;
}

Here is the call graph for this function:

Here is the caller graph for this function:

static struct letterProb* letterProbNew	(	char	letter,
		double	prob
	)			[static, read]

Definition at line 309 of file dnaMotif.c.

References AllocVar, letterProb::letter, and letterProb::prob.

Referenced by addBaseProb().

{
struct letterProb *lp;
AllocVar(lp);
lp->letter = letter;
lp->prob = prob;
return lp;
}

Here is the caller graph for this function:

static void printProbRow	(	FILE *	f,
		char *	label,
		float *	p,
		int	pCount
	)			[static]

Definition at line 257 of file dnaMotif.c.

Referenced by dnaMotifPrintProb().

{
int i;
fprintf(f, "%s ", label);
for (i=0; i < pCount; ++i)
    fprintf(f, "%5.2f ", p[i]);
printf("\n");
}

Here is the caller graph for this function:

static void psOneColumn	(	struct dnaMotif *	motif,
		int	pos,
		double	xStart,
		double	yStart,
		double	width,
		double	totalHeight,
		FILE *	f
	)			[static]

Definition at line 355 of file dnaMotif.c.

References letterProb::letter, letterProbFromMotifColumn(), letterProb::next, letterProb::prob, and slFreeList().

Referenced by dnaMotifToLogoPs(), and dnaMotifToLogoPsW().

{
struct letterProb *lp, *lpList = letterProbFromMotifColumn(motif, pos);
double x = xStart, y = yStart, w = width, h;
for (lp = lpList; lp != NULL; lp = lp->next)
    {
    h = totalHeight * lp->prob;
    if (h >= 1.0)
        {
        fprintf(f, "%cColor ", tolower(lp->letter));
        fprintf(f, "%3.2f ", x);
        fprintf(f, "%3.2f ", y);
        fprintf(f, "%3.2f ", x + w);
        fprintf(f, "%3.2f ", y + h);
        fprintf(f, "(%c) textInBox\n", lp->letter);
        }
    y += h;
    }
fprintf(f, "\n");
slFreeList(&lpList);
}

Here is the call graph for this function:

Here is the caller graph for this function:

static double u1

(

double

prob

)

[static]

Definition at line 277 of file dnaMotif.c.

References logBase2.

Referenced by uncertainty().

{
if (prob == 0)
    return 0;
return prob * logBase2(prob);
}

Here is the caller graph for this function:

static double uncertainty	(	struct dnaMotif *	motif,
		int	pos
	)			[static]

Definition at line 285 of file dnaMotif.c.

References dnaMotif::aProb, dnaMotif::cProb, dnaMotif::gProb, dnaMotif::tProb, and u1().

Referenced by dnaMotifBitsOfInfo().

{
return -( u1(motif->aProb[pos]) + u1(motif->cProb[pos])
        + u1(motif->gProb[pos]) +u1(motif->tProb[pos]) );
}

Here is the call graph for this function:

Here is the caller graph for this function:

---

Variable Documentation

char const rcsid[] = "$Id: dnaMotif.c,v 1.4 2006/09/14 21:08:24 braney Exp $" [static]

Definition at line 12 of file dnaMotif.c.

---