Segment FPR function

eaasna · May 8, 2024 · cb4abb9 · cb4abb9
1 parent 04288ab
commit cb4abb9
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 3 deletions.
diff --git a/include/utilities/threshold/filtering_request.hpp b/include/utilities/threshold/filtering_request.hpp
@@ -41,8 +41,7 @@ struct filtering_request
     {
         double pattern_p = ref_meta.pattern_spurious_match_prob(params);
         uint64_t patterns_per_segment = std::round((query_meta.total_len / (double) query_meta.seg_count - pattern.l + 1) / (double) query_every);
-        double none_match_p = pow(1 - pattern_p, patterns_per_segment);
-        return std::min(1 - none_match_p, 1.0);
+        return segment_fpr(pattern_p, patterns_per_segment);
     }
 
 };

diff --git a/include/utilities/threshold/find.hpp b/include/utilities/threshold/find.hpp
@@ -8,6 +8,15 @@
 namespace valik
 {
 
+/**
+ * @brief The false positive probability of a query segment that contains partially overlapping patterns.
+*/
+inline double segment_fpr(double const pattern_p, size_t const patterns_per_segment)
+{
+    double none_match_p = pow(1 - pattern_p, patterns_per_segment);
+    return std::min(1 - none_match_p, 1.0);
+}
+
 /**
 * @brief Score of the objective function for a parameter set. Smaller values are better.
 */

diff --git a/src/threshold/find.cpp b/src/threshold/find.cpp
@@ -111,7 +111,7 @@ search_kmer_profile find_thresholds_for_kmer_size(metadata const & ref_meta,
         auto best_params = param_set(attr.k, kmer_lemma_threshold(pattern.l, attr.k, errors), space);
 
         if ((best_params.t < THRESH_LOWER) ||  
-        (1 - pow(1 - ref_meta.pattern_spurious_match_prob(best_params), PATTERNS_PER_SEGMENT)) > FPR_UPPER)
+        segment_fpr(ref_meta.pattern_spurious_match_prob(best_params), PATTERNS_PER_SEGMENT) > FPR_UPPER)
         {
             search_type = search_kind::HEURISTIC;
             double best_score = pattern.l;