inc/dnaMotif.h File Reference

#include "dnautil.h"

Include dependency graph for dnaMotif.h:

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

Go to the source code of this file.

Data Structures

struct  dnaMotif

Defines

#define dnaMotifTabOut(el, f)   dnaMotifOutput(el,f,'\t','\n');
#define dnaMotifCommaOut(el, f)   dnaMotifOutput(el,f,',',',');

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)
double dnaMotifBitsOfInfo (struct dnaMotif *motif, int pos)
void dnaMotifPrintProb (struct dnaMotif *motif, FILE *f)
void dnaMotifToLogoPs (struct dnaMotif *motif, double widthPerBase, double height, char *fileName)
void dnaMotifToLogoPsW (struct dnaMotif *motif, double widthPerBase, double width, double height, char *fileName)
void dnaMotifToLogoPng (struct dnaMotif *motif, double widthPerBase, double height, char *gsExe, char *tempDir, char *fileName)
void dnaMotifToLogoPGM (struct dnaMotif *motif, double widthPerBase, double width, double height, char *gsExe, char *tempDir, char *fileName)

Define Documentation

#define dnaMotifCommaOut ( el,
f   )     dnaMotifOutput(el,f,',',',');

Definition at line 42 of file dnaMotif.h.

#define dnaMotifTabOut ( el,
f   )     dnaMotifOutput(el,f,'\t','\n');

Definition at line 39 of file dnaMotif.h.

Function Documentation

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

Definition at line 173 of file dnaMotif.c.

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

00175 {
00176 float fScore, rScore;
00177 fScore = dnaMotifSequenceProb(motif, dna);
00178 reverseComplement(dna, motif->columnCount);
00179 rScore = dnaMotifSequenceProb(motif, dna);
00180 reverseComplement(dna, motif->columnCount);
00181 if (fScore >= rScore)
00182     return '+';
00183 else
00184     return '-';
00185 }

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.

00189 {
00190 double p = dnaMotifSequenceProb(motif, dna);
00191 double q = pow(0.25, motif->columnCount);
00192 double odds = p/q;
00193 return logBase2(odds);
00194 }

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().

00295 {
00296 if (pos > motif->columnCount || pos < 0)
00297     internalErr();
00298 return 2 - uncertainty(motif, pos);
00299 }

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().

00018 {
00019 char *s = *pS;
00020 int i;
00021 
00022 if (ret == NULL)
00023     AllocVar(ret);
00024 ret->name = sqlStringComma(&s);
00025 ret->columnCount = sqlSignedComma(&s);
00026 s = sqlEatChar(s, '{');
00027 AllocArray(ret->aProb, ret->columnCount);
00028 for (i=0; i<ret->columnCount; ++i)
00029     {
00030     ret->aProb[i] = sqlSignedComma(&s);
00031     }
00032 s = sqlEatChar(s, '}');
00033 s = sqlEatChar(s, ',');
00034 s = sqlEatChar(s, '{');
00035 AllocArray(ret->cProb, ret->columnCount);
00036 for (i=0; i<ret->columnCount; ++i)
00037     {
00038     ret->cProb[i] = sqlSignedComma(&s);
00039     }
00040 s = sqlEatChar(s, '}');
00041 s = sqlEatChar(s, ',');
00042 s = sqlEatChar(s, '{');
00043 AllocArray(ret->gProb, ret->columnCount);
00044 for (i=0; i<ret->columnCount; ++i)
00045     {
00046     ret->gProb[i] = sqlSignedComma(&s);
00047     }
00048 s = sqlEatChar(s, '}');
00049 s = sqlEatChar(s, ',');
00050 s = sqlEatChar(s, '{');
00051 AllocArray(ret->tProb, ret->columnCount);
00052 for (i=0; i<ret->columnCount; ++i)
00053     {
00054     ret->tProb[i] = sqlSignedComma(&s);
00055     }
00056 s = sqlEatChar(s, '}');
00057 s = sqlEatChar(s, ',');
00058 *pS = s;
00059 return ret;
00060 }

Here is the call 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().

00065 {
00066 struct dnaMotif *el;
00067 
00068 if ((el = *pEl) == NULL) return;
00069 freeMem(el->name);
00070 freeMem(el->aProb);
00071 freeMem(el->cProb);
00072 freeMem(el->gProb);
00073 freeMem(el->tProb);
00074 freez(pEl);
00075 }

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.

00079 {
00080 struct dnaMotif *el, *next;
00081 
00082 for (el = *pList; el != NULL; el = next)
00083     {
00084     next = el->next;
00085     dnaMotifFree(&el);
00086     }
00087 *pList = NULL;
00088 }

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().

00219 {
00220 int i;
00221 for (i=0; i<motif->columnCount; ++i)
00222     {
00223     if (motif->aProb[i] < 0) return TRUE;
00224     if (motif->cProb[i] < 0) return TRUE;
00225     if (motif->gProb[i] < 0) return TRUE;
00226     if (motif->tProb[i] < 0) return TRUE;
00227     }
00228 return FALSE;
00229 }

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().

00250 {
00251 if (dnaMotifIsScoreBased(motif))
00252     dnaMotifScoreToProb(motif);
00253 else
00254     dnaMotifNormalize(motif);
00255 }

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().

00199 {
00200 int i;
00201 for (i=0; i<motif->columnCount; ++i)
00202     {
00203     float sum = motif->aProb[i] + motif->cProb[i] + motif->gProb[i] + motif->tProb[i];
00204     if (sum < 0)
00205         errAbort("%s has negative numbers, perhaps it's score not probability based", 
00206                 motif->name);
00207     if (sum == 0)
00208          motif->aProb[i] = motif->cProb[i] = motif->gProb[i] = motif->tProb[i] = 0.25;
00209     motif->aProb[i] /= sum;
00210     motif->cProb[i] /= sum;
00211     motif->gProb[i] /= sum;
00212     motif->tProb[i] /= sum;
00213     }
00214 }

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.

00092 {
00093 int i;
00094 if (sep == ',') fputc('"',f);
00095 fprintf(f, "%s", el->name);
00096 if (sep == ',') fputc('"',f);
00097 fputc(sep,f);
00098 fprintf(f, "%d", el->columnCount);
00099 fputc(sep,f);
00100 if (sep == ',') fputc('{',f);
00101 for (i=0; i<el->columnCount; ++i)
00102     {
00103     fprintf(f, "%f", el->aProb[i]);
00104     fputc(',', f);
00105     }
00106 if (sep == ',') fputc('}',f);
00107 fputc(sep,f);
00108 if (sep == ',') fputc('{',f);
00109 for (i=0; i<el->columnCount; ++i)
00110     {
00111     fprintf(f, "%f", el->cProb[i]);
00112     fputc(',', f);
00113     }
00114 if (sep == ',') fputc('}',f);
00115 fputc(sep,f);
00116 if (sep == ',') fputc('{',f);
00117 for (i=0; i<el->columnCount; ++i)
00118     {
00119     fprintf(f, "%f", el->gProb[i]);
00120     fputc(',', f);
00121     }
00122 if (sep == ',') fputc('}',f);
00123 fputc(sep,f);
00124 if (sep == ',') fputc('{',f);
00125 for (i=0; i<el->columnCount; ++i)
00126     {
00127     fprintf(f, "%f", el->tProb[i]);
00128     fputc(',', f);
00129     }
00130 if (sep == ',') fputc('}',f);
00131 fputc(lastSep,f);
00132 }

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.

00269 {
00270 printProbRow(f, "A", motif->aProb, motif->columnCount);
00271 printProbRow(f, "C", motif->cProb, motif->columnCount);
00272 printProbRow(f, "G", motif->gProb, motif->columnCount);
00273 printProbRow(f, "T", motif->tProb, motif->columnCount);
00274 }

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().

00234                                     : 25% for each base */
00235 {
00236 int i;
00237 for (i=0; i<motif->columnCount; ++i)
00238     {
00239     motif->aProb[i] = exp(motif->aProb[i]);
00240     motif->cProb[i] = exp(motif->cProb[i]);
00241     motif->gProb[i] = exp(motif->gProb[i]);
00242     motif->tProb[i] = exp(motif->tProb[i]);
00243     }
00244 dnaMotifNormalize(motif);
00245 }

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().

00138 {
00139 float p = 1.0;
00140 int i;
00141 for (i=0; i<motif->columnCount; ++i)
00142     {
00143     switch (dna[i])
00144         {
00145         case 'a':
00146         case 'A':
00147             p *= motif->aProb[i];
00148             break;
00149         case 'c':
00150         case 'C':
00151             p *= motif->cProb[i];
00152             break;
00153         case 'g':
00154         case 'G':
00155             p *= motif->gProb[i];
00156             break;
00157         case 't':
00158         case 'T':
00159             p *= motif->tProb[i];
00160             break;
00161         case 0:
00162             warn("dna shorter than motif");
00163             internalErr();
00164             break;
00165         default:
00166             p *= 0.25;
00167             break;
00168         }
00169     }
00170 return p;
00171 }

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.

00493 {
00494 char *psName = rTempName(tempDir, "dnaMotif", ".ps");
00495 struct dyString *dy = dyStringNew(0);
00496 int w, h;
00497 int sysRet;
00498 
00499 if (gsExe == NULL) gsExe = "gs";
00500 if (tempDir == NULL) tempDir = "/tmp";
00501 dnaMotifToLogoPsW(motif, widthPerBase, width, height, psName);
00502 dnaMotifDims(motif, widthPerBase, height, &w, &h);
00503 dyStringAppend(dy, gsExe);
00504 dyStringAppend(dy, " -sDEVICE=pgmraw -sOutputFile=");
00505 dyStringAppend(dy, fileName);
00506 dyStringAppend(dy, " -dBATCH -dNOPAUSE -q ");
00507 dyStringPrintf(dy, "-g%dx%d ", (int) ceil(width), h);
00508 dyStringAppend(dy, psName);
00509 sysRet = system(dy->string);
00510 if (sysRet != 0)
00511     errAbort("System call returned %d for:\n  %s", sysRet, dy->string);
00512 
00513 /* Clean up. */
00514 dyStringFree(&dy);
00515 
00516 /* change permisssions so the webserver can access the file */
00517 dy = newDyString(0);
00518 dyStringPrintf(dy, "chmod 666 %s ", fileName);
00519 sysRet = system(dy->string);
00520 
00521 remove(psName);
00522 }

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.

00425 {
00426 char *psName = rTempName(tempDir, "dnaMotif", ".ps");
00427 struct dyString *dy = dyStringNew(0);
00428 int w, h;
00429 int sysRet;
00430 
00431 if (gsExe == NULL) gsExe = "gs";
00432 if (tempDir == NULL) tempDir = "/tmp";
00433 dnaMotifToLogoPs(motif, widthPerBase, height, psName);
00434 dnaMotifDims(motif, widthPerBase, height, &w, &h);
00435 dyStringAppend(dy, gsExe);
00436 dyStringAppend(dy, " -sDEVICE=png16m -sOutputFile=");
00437 dyStringAppend(dy, fileName);
00438 dyStringAppend(dy, " -dBATCH -dNOPAUSE -q ");
00439 dyStringPrintf(dy, "-g%dx%d ", w, h);
00440 dyStringAppend(dy, psName);
00441 sysRet = system(dy->string);
00442 if (sysRet != 0)
00443     errAbort("System call returned %d for:\n  %s", sysRet, dy->string);
00444 
00445 /* Clean up. */
00446 dyStringFree(&dy);
00447 
00448 /* change permisssions so the webserver can access the file */
00449 dy = newDyString(0);
00450 dyStringPrintf(dy, "chmod 666 %s ", fileName);
00451 sysRet = system(dy->string);
00452 
00453 remove(psName);
00454 }

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().

00391 {
00392 FILE *f = mustOpen(fileName, "w");
00393 int i;
00394 int xStart = 0;
00395 int w, h;
00396 char *s = 
00397 #include "dnaMotif.pss"
00398 ;
00399 
00400 dnaMotifDims(motif, widthPerBase, height, &w, &h);
00401 fprintf(f, "%%!PS-Adobe-3.1 EPSF-3.0\n");
00402 fprintf(f, "%%%%BoundingBox: 0 0 %d %d\n\n", w, h);
00403 fprintf(f, "%s", s);
00404 
00405 fprintf(f, "%s", "% Start of code for this specific logo\n");
00406 
00407 for (i=0; i<motif->columnCount; ++i)
00408     {
00409     double infoScale = dnaMotifBitsOfInfo(motif, i)/2.0;
00410     psOneColumn(motif, i, xStart, 0, widthPerBase, infoScale * height, f);
00411     xStart += widthPerBase;
00412     }
00413 fprintf(f, "showpage\n");
00414 carefulClose(&f);
00415 }

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().

00458 {
00459 FILE *f = mustOpen(fileName, "w");
00460 int i;
00461 int xStart = 0;
00462 int w, h;
00463 char *s = 
00464 #include "dnaMotif.pss"
00465 ;
00466 
00467 dnaMotifDims(motif, widthPerBase, height, &w, &h);
00468 fprintf(f, "%%!PS-Adobe-3.1 EPSF-3.0\n");
00469 fprintf(f, "%%%%BoundingBox: 0 0 %d %d\n\n", w, h);
00470 fprintf(f, "%s", s);
00471 
00472 fprintf(f, "%s", "% Start of code for this specific logo\n");
00473 
00474 for (i=0; i<motif->columnCount; ++i)
00475     {
00476     double infoScale = 0.9 ; 
00477     xStart = i * width / motif->columnCount;
00478     psOneColumn(motif, i, xStart, 0, widthPerBase, infoScale * height, f);
00479     }
00480 fprintf(f, "showpage\n");
00481 carefulClose(&f);
00482 }

Here is the call graph for this function:

Here is the caller graph for this function:

Generated on Tue Dec 25 18:51:55 2007 for blat by  doxygen 1.5.2