From 61acb788f682e8a26a22b94d65b49b07d7dc3498 Mon Sep 17 00:00:00 2001
From: Tobias Schmidt <tobias.schmidt@unibas.ch>
Date: Wed, 13 Jul 2011 14:18:25 +0200
Subject: [PATCH] fixes weight matrix problem for non-alpha characters
(BZDNG-280)
---
modules/seq/alg/src/subst_weight_matrix.hh | 31 +++++++++++++++++-----
modules/seq/alg/tests/CMakeLists.txt | 1 +
2 files changed, 26 insertions(+), 6 deletions(-)
diff --git a/modules/seq/alg/src/subst_weight_matrix.hh b/modules/seq/alg/src/subst_weight_matrix.hh
index cdf0fc83a..f310893ab 100644
--- a/modules/seq/alg/src/subst_weight_matrix.hh
+++ b/modules/seq/alg/src/subst_weight_matrix.hh
@@ -47,24 +47,43 @@ public:
::memset(weights_, 0, sizeof(WeightType)*ALPHABET_SIZE*ALPHABET_SIZE);
}
+ /// \brief Get the substitution weight between two amino acids
+ ///
+ /// If the amino acid single letter code is unknown (e.g. '?')
+ /// a weight of 0 is returned
WeightType GetWeight(char aa_one, char aa_two) const
{
+ if (!(IsAlpha(aa_one) && IsAlpha(aa_two))) {
+ return 0;
+ }
int i=Index(aa_one, aa_two);
- assert(i>=0 && i<ALPHABET_SIZE*ALPHABET_SIZE);
- return weights_[i];
+ return (i>=0 && i<ALPHABET_SIZE*ALPHABET_SIZE) ? weights_[i] : 0;
}
+ /// \brief Set the substitution weight between two amino acids
+ ///
+ /// The weight is only set if the amino acid single letter code
+ /// is known (e.g. no weight is set for '?')
void SetWeight(char aa_one, char aa_two, WeightType weight)
{
- int i=Index(aa_one, aa_two);
- assert(i>=0 && i<ALPHABET_SIZE*ALPHABET_SIZE);
- weights_[i]=weight;
+ if ((IsAlpha(aa_one) && IsAlpha(aa_two))) {
+ int i=Index(aa_one, aa_two);
+ if (i>=0 && i<ALPHABET_SIZE*ALPHABET_SIZE) {
+ weights_[i]=weight;
+ }
+ }
}
+
private:
int Index(char aa_one, char aa_two) const {
return (toupper(aa_one)-'A')*ALPHABET_SIZE+(toupper(aa_two)-'A');
}
- WeightType weights_[ALPHABET_SIZE*ALPHABET_SIZE];
+
+ /// \brief Check if uppercase character is one of [A-Z]
+ bool IsAlpha(char aa) const {
+ return (toupper(aa)>='A' && toupper(aa)<='Z');
+ }
+ WeightType weights_[ALPHABET_SIZE*ALPHABET_SIZE];
};
SubstWeightMatrixPtr DLLEXPORT_OST_SEQ_ALG
diff --git a/modules/seq/alg/tests/CMakeLists.txt b/modules/seq/alg/tests/CMakeLists.txt
index 993352440..3a26d1d67 100644
--- a/modules/seq/alg/tests/CMakeLists.txt
+++ b/modules/seq/alg/tests/CMakeLists.txt
@@ -5,6 +5,7 @@ set(OST_SEQ_ALG_UNIT_TESTS
test_renumber.py
test_local_align.py
test_global_align.py
+ test_weight_matrix.py
)
ost_unittest(MODULE seq_alg SOURCES "${OST_SEQ_ALG_UNIT_TESTS}")
--
GitLab