Merge pull request #53 from jaebeom-kim/master

new parameter: --tie-ratio
steineggerlab · Jan 23, 2024 · cdff8d4 · cdff8d4
2 parents 0b60d81 + f2c33e8
commit cdff8d4
Show file tree

Hide file tree

Showing 5 changed files with 15 additions and 2 deletions.
diff --git a/src/commons/LocalParameters.cpp b/src/commons/LocalParameters.cpp
@@ -150,6 +150,13 @@ LocalParameters::LocalParameters() :
  typeid(int),
  (void *) &minSSMatch,
  "^[0-9]+$"),
+ TIE_RATIO(TIE_RATIO_ID,
+ "--tie-ratio",
+ "Best * --tie-ratio is considered as a tie",
+ "Best * --tie-ratio is considered as a tie",
+ typeid(float),
+ (void *) &tieRatio,
+ "^0(\\.[0-9]+)?|1(\\.0+)?$"),
  LIBRARY_PATH(LIBRARY_PATH_ID,
  "--library-path",
  "Path to library where the FASTA files are stored",
@@ -314,6 +321,7 @@ LocalParameters::LocalParameters() :
  classify.push_back(&RAM_USAGE);
  classify.push_back(&MATCH_PER_KMER);
  classify.push_back(&ACCESSION_LEVEL);
+ classify.push_back(&TIE_RATIO);
  // classify.push_back(&MIN_SS_MATCH);
 
  // filter 

diff --git a/src/commons/LocalParameters.h b/src/commons/LocalParameters.h
@@ -56,6 +56,7 @@ class LocalParameters : public Parameters {
  PARAMETER(MIN_CONS_CNT_EUK)
  PARAMETER(MATCH_PER_KMER)
  PARAMETER(MIN_SS_MATCH)
+ PARAMETER(TIE_RATIO)
 
  // DB build parameters
  PARAMETER(LIBRARY_PATH)
@@ -104,6 +105,7 @@ class LocalParameters : public Parameters {
  int minConsCntEuk;
  int matchPerKmer;
  int minSSMatch;
+ float tieRatio;
 
  // Database creation
  std::string tinfoPath;

diff --git a/src/commons/Taxonomer.cpp b/src/commons/Taxonomer.cpp
@@ -27,6 +27,7 @@ Taxonomer::Taxonomer(const LocalParameters &par, NcbiTaxonomy *taxonomy) : taxon
  minConsCnt = par.minConsCnt;
  minConsCntEuk = par.minConsCntEuk;
  eukaryotaTaxId = par.eukaryotaTaxId;
+ tieRatio = par.tieRatio;
 
  if (par.seqMode == 1 || par.seqMode == 2) {
  denominator = 100;
@@ -300,7 +301,7 @@ TaxonScore Taxonomer::getBestSpeciesMatches(vector<Match> & speciesMatches,
 
  vector<TaxID> maxSpecies;
  for (auto & spScore : species2score) {
- if (spScore.second > bestSpScore * 0.95) {
+ if (spScore.second >= bestSpScore * tieRatio) {
  maxSpecies.push_back(spScore.first);
  }
  }
@@ -397,7 +398,7 @@ TaxonScore Taxonomer::getBestSpeciesMatches(vector<Match> & speciesMatches,
  }
  vector<TaxID> maxSpecies;
  for (auto & spScore : species2score) {
- if (spScore.second > bestSpScore * 0.95) {
+ if (spScore.second >= bestSpScore * tieRatio) {
  maxSpecies.push_back(spScore.first);
  }
  }

diff --git a/src/commons/Taxonomer.h b/src/commons/Taxonomer.h
@@ -61,6 +61,7 @@ class Taxonomer {
  int minConsCnt;
  int minConsCntEuk;
  int eukaryotaTaxId;
+ float tieRatio;
 
  // Internal
  int denominator;

diff --git a/src/workflow/classify.cpp b/src/workflow/classify.cpp
@@ -25,6 +25,7 @@ void setClassifyDefaults(LocalParameters & par){
  par.maskProb = 0.9;
  par.matchPerKmer = 4;
  par.accessionLevel = 0;
+ par.tieRatio = 0.95;
 }
 
 int classify(int argc, const char **argv, const Command& command)