[med-svn] [Git][med-team/tantan][upstream] New upstream version 49
Sascha Steinbiss (@satta)
gitlab at salsa.debian.org
Sat Dec 23 22:06:44 GMT 2023
Sascha Steinbiss pushed to branch upstream at Debian Med / tantan
Commits:
6cf87cf6 by Sascha Steinbiss at 2023-12-23T23:03:16+01:00
New upstream version 49
- - - - -
5 changed files:
- src/Makefile
- src/tantan_repeat_finder.cc
- src/tantan_repeat_finder.hh
- test/tantan_test.out
- test/tantan_test.sh
Changes:
=====================================
src/Makefile
=====================================
@@ -10,7 +10,7 @@ clean:
rm -f ../bin/tantan
VERSION1 = git describe --dirty
-VERSION2 = echo ' (HEAD -> main, tag: 40) ' | sed -e 's/.*tag: *//' -e 's/[,) ].*//'
+VERSION2 = echo ' (HEAD -> main, tag: 49) ' | sed -e 's/.*tag: *//' -e 's/[,) ].*//'
VERSION = \"`test -e ../.git && $(VERSION1) || $(VERSION2)`\"
=====================================
src/tantan_repeat_finder.cc
=====================================
@@ -71,9 +71,14 @@ void RepeatFinder::init(int maxRepeatOffset,
f2f1 = myLog(1 - repeatEndProb - firstGapProb);
f2f2 = myLog(1 - repeatEndProb - firstGapProb * 2);
- double x = 1 / repeatOffsetProbDecay;
- b2fGrowth = myLog(x);
- b2fLast = myLog(repeatProb * firstRepeatOffsetProb(x, maxRepeatOffset));
+ double b2fDecay = myLog(repeatOffsetProbDecay);
+ double b2fFirst = myLog(firstRepeatOffsetProb(repeatOffsetProbDecay,
+ maxRepeatOffset) * repeatProb);
+
+ b2fScores.resize(maxRepeatOffset + 1);
+ for (int i = 0; i < maxRepeatOffset; ++i) {
+ b2fScores[i+1] = b2fFirst + i * b2fDecay;
+ }
dpScoresPerLetter = numOfDpScoresPerLetter(maxRepeatOffset, endGapScore);
}
@@ -86,48 +91,16 @@ void RepeatFinder::initializeBackwardScores() {
}
}
-void RepeatFinder::calcBackwardTransitionScoresWithGaps() {
- double toBackground = f2b + scoresPtr[0];
- double *foregroundPtr = scoresPtr + 1;
- double f = *foregroundPtr;
- double toForeground = f;
-
- double *insertionPtr = scoresPtr + 1 + maxRepeatOffset;
- double i = *insertionPtr;
- *foregroundPtr = max3(toBackground, f2f1 + f, i);
- double d = endGapScore + f;
- ++foregroundPtr;
- toForeground += b2fGrowth;
-
- while (foregroundPtr < scoresPtr + maxRepeatOffset) {
- f = *foregroundPtr;
- toForeground = std::max(toForeground, f);
- i = *(insertionPtr + 1);
- *foregroundPtr = max3(toBackground, f2f2 + f, std::max(i, d));
- double oneGapScore_f = oneGapScore + f;
- *insertionPtr = std::max(oneGapScore_f, g2g + i);
- d = std::max(oneGapScore_f, g2g + d);
- ++foregroundPtr;
- ++insertionPtr;
- toForeground += b2fGrowth;
- }
-
- f = *foregroundPtr;
- toForeground = std::max(toForeground, f);
- *foregroundPtr = max3(toBackground, f2f1 + f, d);
- *insertionPtr = endGapScore + f;
-
- scoresPtr[0] = std::max(b2b + scoresPtr[0], b2fLast + toForeground);
-}
-
-void RepeatFinder::calcEmissionScores() {
+void RepeatFinder::calcScoresForOneSequencePositionWithGaps() {
+ const double *b2f = &b2fScores[0];
+ const double *oldScores = scoresPtr - dpScoresPerLetter;
+ const double *oldInsertionScores = oldScores + maxRepeatOffset;
const double *matrixRow = substitutionMatrix[*seqPtr];
- double *oldScores = scoresPtr - dpScoresPerLetter;
+ double *newInsertionScores = scoresPtr + maxRepeatOffset - 1;
+ double toBackground = f2b + oldScores[0];
int maxOffset = maxOffsetInTheSequence();
int i = 1;
- scoresPtr[0] = oldScores[0];
-
for (; i <= maxOffset; ++i) {
scoresPtr[i] = oldScores[i] + matrixRow[seqPtr[-i]];
}
@@ -136,37 +109,56 @@ void RepeatFinder::calcEmissionScores() {
scoresPtr[i] = -HUGE_VAL;
}
- std::copy(oldScores + i, scoresPtr, scoresPtr + i);
+ i = 1;
+ double f = scoresPtr[i];
+ double toForeground = f + b2f[i];
+ double ins = oldInsertionScores[i];
+ scoresPtr[i] = max3(toBackground, f2f1 + f, ins);
+ double del = endGapScore + f;
+
+ for (i = 2; i < maxRepeatOffset; ++i) {
+ f = scoresPtr[i];
+ toForeground = std::max(toForeground, f + b2f[i]);
+ ins = oldInsertionScores[i];
+ scoresPtr[i] = max3(toBackground, f2f2 + f, std::max(ins, del));
+ double oneGapScore_f = oneGapScore + f;
+ newInsertionScores[i] = std::max(oneGapScore_f, g2g + ins);
+ del = std::max(oneGapScore_f, g2g + del);
+ }
+
+ f = scoresPtr[i];
+ toForeground = std::max(toForeground, f + b2f[i]);
+ scoresPtr[i] = max3(toBackground, f2f1 + f, del);
+ newInsertionScores[i] = endGapScore + f;
+
+ scoresPtr[0] = std::max(b2b + oldScores[0], toForeground);
}
void RepeatFinder::calcScoresForOneSequencePosition() {
if (endGapScore > -HUGE_VAL) {
- calcEmissionScores();
- calcBackwardTransitionScoresWithGaps();
+ calcScoresForOneSequencePositionWithGaps();
return;
}
- double *oldScores = scoresPtr - dpScoresPerLetter;
- double toBackground = f2b + oldScores[0];
+ const double *b2f = &b2fScores[0];
+ const double *oldScores = scoresPtr - dpScoresPerLetter;
const double *matrixRow = substitutionMatrix[*seqPtr];
+ double toBackground = f2b + oldScores[0];
int maxOffset = maxOffsetInTheSequence();
double toForeground = -HUGE_VAL;
int i = 1;
for (; i <= maxOffset; ++i) {
double f = oldScores[i] + matrixRow[seqPtr[-i]];
- toForeground = std::max(toForeground + b2fGrowth, f);
+ toForeground = std::max(toForeground, f + b2f[i]);
scoresPtr[i] = std::max(toBackground, f2f0 + f);
}
for (; i <= maxRepeatOffset; ++i) {
- toForeground += b2fGrowth;
scoresPtr[i] = toBackground;
}
- std::copy(oldScores + i, scoresPtr, scoresPtr + i);
-
- scoresPtr[0] = std::max(b2b + oldScores[0], b2fLast + toForeground);
+ scoresPtr[0] = std::max(b2b + oldScores[0], toForeground);
}
void RepeatFinder::makeCheckpoint() {
@@ -222,8 +214,7 @@ int RepeatFinder::offsetWithMaxScore() const {
double toForeground = -HUGE_VAL;
for (int i = 1; i <= maxOffset; ++i) {
- toForeground += b2fGrowth;
- double f = scoreWithEmission(matrixRow, i);
+ double f = scoreWithEmission(matrixRow, i) + b2fScores[i];
if (f > toForeground) {
toForeground = f;
bestOffset = i;
=====================================
src/tantan_repeat_finder.hh
=====================================
@@ -51,8 +51,8 @@ private:
double f2f0;
double f2f1;
double f2f2;
- double b2fGrowth;
- double b2fLast;
+
+ std::vector<double> b2fScores;
int maxRepeatOffset;
int dpScoresPerLetter;
@@ -68,8 +68,7 @@ private:
int state;
void initializeBackwardScores();
- void calcBackwardTransitionScoresWithGaps();
- void calcEmissionScores();
+ void calcScoresForOneSequencePositionWithGaps();
void calcScoresForOneSequencePosition();
void makeCheckpoint();
void redoCheckpoint();
=====================================
test/tantan_test.out
=====================================
@@ -1533,3 +1533,5 @@ SRR019778.65 2 13 2 5.5 TG TG,TG,TG,TG,TG,T
SRR019778.78 9 31 10 2.2 TGCCTTACTA TGCCTTACTA,TGCCTTACTA,TG
SRR019778.95 2 18 4 4 TGAT TGAT,TGAT,TGAT,TGAT
SRR019778.95 22 45 4 5.75 ATAG ATAG,ATAG,ATAG,ATAG,ATAG,ATA
+
+hard 0 40 5 7.6 ATCAT CATCAT,CATCAT,CAT-AT,CATAT,CATAT,CATAT,CATAT,CAT
=====================================
test/tantan_test.sh
=====================================
@@ -39,4 +39,6 @@ countLowercaseLetters () {
tantan -f4 -b0 panda.fastq
echo
tantan -f4 -b0 -j0 panda.fastq
+ echo
+ tantan -f4 -w7000 hard.fa
} 2>&1 | diff -u tantan_test.out -
View it on GitLab: https://salsa.debian.org/med-team/tantan/-/commit/6cf87cf68cee77560671714d4772ac9d9810a368
--
View it on GitLab: https://salsa.debian.org/med-team/tantan/-/commit/6cf87cf68cee77560671714d4772ac9d9810a368
You're receiving this email because of your account on salsa.debian.org.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://alioth-lists.debian.net/pipermail/debian-med-commit/attachments/20231223/49c4a64b/attachment-0001.htm>
More information about the debian-med-commit
mailing list