From 24927ca428e844b92b8b570d2a72acaa04ab10c5 Mon Sep 17 00:00:00 2001
From: Stefan Bienert <stefan.bienert@unibas.ch>
Date: Fri, 27 Sep 2013 11:34:27 +0200
Subject: [PATCH] Added function 'Normalise()' for sequences
---
modules/seq/base/doc/seq.rst | 11 +++++++----
modules/seq/base/pymod/export_sequence.cc | 1 +
modules/seq/base/src/impl/sequence_impl.cc | 21 +++++++++++++++++++++
modules/seq/base/src/impl/sequence_impl.hh | 3 +++
modules/seq/base/src/sequence_handle.cc | 5 +++++
modules/seq/base/src/sequence_handle.hh | 8 ++++++--
modules/seq/base/tests/test_seq.py | 10 ++++++++++
modules/seq/base/tests/test_sequence.cc | 20 ++++++++++++++++++++
8 files changed, 73 insertions(+), 6 deletions(-)
diff --git a/modules/seq/base/doc/seq.rst b/modules/seq/base/doc/seq.rst
index 042f1d09c..e5a89a5f1 100644
--- a/modules/seq/base/doc/seq.rst
+++ b/modules/seq/base/doc/seq.rst
@@ -30,7 +30,7 @@ Sequence Offset
When using sequences and structures together, often the start of the structure
and the beginning of the sequence do not fall together. In the following case,
-the alignment of sequences B and C only covers a subpart of structure A::
+the alignment of sequences B and C only covers a subsequence of structure A::
A acefghiklmnpqrstuvwy
B ghiklm
@@ -167,6 +167,10 @@ The SequenceHandle
Returns a string version of this sequence with all hyphens removed. Also
available as the property :attr:`gapless_string`.
+ .. method:: Normalise()
+
+ Remove '-' and '.' as gaps from the sequence and make it all upper case.
+ Works in place.
.. method:: SetName()
@@ -205,7 +209,7 @@ The SequenceHandle
Check whether the two sequences s1 and s2 match. This function performs are
case-insensitive comparison of the two sequences. The character 'X' is
- interpreted as a wildcard character that always matches the other sequence.
+ interpreted as a wild card character that always matches the other sequence.
The SequenceList
--------------------------------------------------------------------------------
@@ -220,7 +224,7 @@ The AlignmentHandle
--------------------------------------------------------------------------------
The :class:`AlignmentHandle` represents a list of aligned sequences. In
-constrast to :class:`SequenceList`, an alignment requires all sequences to be of
+contrast to :class:`SequenceList`, an alignment requires all sequences to be of
the same length. New instances of alignments are created with
:func:`CreateAlignment` and :func:`AlignmentFromSequenceList`.
@@ -431,4 +435,3 @@ an alignment:
.. method:: RemoveSequence(index)
Remove sequence at *index* from the alignment.
-
diff --git a/modules/seq/base/pymod/export_sequence.cc b/modules/seq/base/pymod/export_sequence.cc
index 6cdf081cd..1919290e7 100644
--- a/modules/seq/base/pymod/export_sequence.cc
+++ b/modules/seq/base/pymod/export_sequence.cc
@@ -300,6 +300,7 @@ void export_sequence()
.def("AttachView", attach_two)
.def("Append", &SequenceHandle::Append)
.def("SetString", &SequenceHandle::SetString)
+ .def("Normalise", &SequenceHandle::Normalise)
.add_property("string",
make_function(&SequenceHandle::GetString,
return_value_policy<copy_const_reference>()),
diff --git a/modules/seq/base/src/impl/sequence_impl.cc b/modules/seq/base/src/impl/sequence_impl.cc
index 733eaf1b8..9864fe8e5 100644
--- a/modules/seq/base/src/impl/sequence_impl.cc
+++ b/modules/seq/base/src/impl/sequence_impl.cc
@@ -331,6 +331,27 @@ void SequenceImpl::Replace(const String& str,int start, int end)
this->ShiftsFromSequence();
}
+void SequenceImpl::Normalise() {
+ const char gaps[]={'.', '-', '\0'};
+ bool is_gap;
+ String new_seq_string = "";
+ int n_gaps = sizeof(gaps) / sizeof(gaps[0]);
+ for (size_t i=0; i<seq_string_.length(); ++i) {
+ is_gap = false;
+ for (int j=0; j<n_gaps; ++j) {
+ if (seq_string_[i]==gaps[j]) {
+ is_gap = true;
+ break;
+ }
+ }
+ if (is_gap == false) {
+ new_seq_string = new_seq_string + char(toupper(seq_string_[i]));
+ }
+ }
+ seq_string_ = new_seq_string;
+ this->ShiftsFromSequence();
+}
+
void SequenceImpl::ShiftRegion(int start, int end, int amount)
{
if(start > end || start + amount < 0 || end + amount > this->GetLength()){
diff --git a/modules/seq/base/src/impl/sequence_impl.hh b/modules/seq/base/src/impl/sequence_impl.hh
index fecad2ca9..ccb64b775 100644
--- a/modules/seq/base/src/impl/sequence_impl.hh
+++ b/modules/seq/base/src/impl/sequence_impl.hh
@@ -81,6 +81,9 @@ public:
/// \brief Set sequence String
void SetString(const String& seq);
+ /// \brief Remove everything 'unusual' from sequence
+ void Normalise();
+
/// \brief replace substring starting from start to end
void Replace(const String& str,int start, int end);
diff --git a/modules/seq/base/src/sequence_handle.cc b/modules/seq/base/src/sequence_handle.cc
index 627ebdc36..f57a624e7 100644
--- a/modules/seq/base/src/sequence_handle.cc
+++ b/modules/seq/base/src/sequence_handle.cc
@@ -295,6 +295,11 @@ String SequenceHandle::GetGaplessString() const
return Impl()->GetGaplessString();
}
+void SequenceHandle::Normalise() {
+ this->CheckValidity();
+ Impl()->Normalise();
+}
+
int SequenceHandle::GetOffset() const
{
this->CheckValidity();
diff --git a/modules/seq/base/src/sequence_handle.hh b/modules/seq/base/src/sequence_handle.hh
index d7247cbd1..cf9aa732d 100644
--- a/modules/seq/base/src/sequence_handle.hh
+++ b/modules/seq/base/src/sequence_handle.hh
@@ -123,7 +123,7 @@ public:
bool HasAttachedView() const;
const String& GetRole() const;
-
+
bool operator==(const ConstSequenceHandle& rhs) const;
bool operator!=(const ConstSequenceHandle& rhs) const;
@@ -231,7 +231,11 @@ public:
/// \brief get attached view. may be an invalid entity view
/// \sa SequenceHandle::AttachView(const mol::EntityView&, const String&)
mol::EntityView GetAttachedView() const;
-
+
+ /// \brief remove '.', '-' as gaps and make sequence all-uppercase
+ /// Changes happen in place.
+ void Normalise();
+
/// \brief create copy sequence
/// The newly created sequence has the same attached view.
SequenceHandle Copy() const;
diff --git a/modules/seq/base/tests/test_seq.py b/modules/seq/base/tests/test_seq.py
index bfd80d7bf..f202fe522 100644
--- a/modules/seq/base/tests/test_seq.py
+++ b/modules/seq/base/tests/test_seq.py
@@ -175,6 +175,16 @@ class TestSeq(unittest.TestCase):
self.assertEqual(string_a, 'BDFH')
self.assertEqual(string_b, 'BDFH')
+ def testNormalise(self):
+ seq_a=seq.CreateSequence("A", "B-D-FGH")
+ self.assertEqual("B-D-FGH", seq_a.GetString())
+ seq_a.Normalise()
+ self.assertEqual("BDFGH", seq_a.GetString())
+ seq_a=seq.CreateSequence("A", "b.d-fgh")
+ self.assertEqual("b.d-fgh", seq_a.GetString())
+ seq_a.Normalise()
+ self.assertEqual("BDFGH", seq_a.GetString())
+
if __name__== '__main__':
from ost import testutils
testutils.RunTests()
diff --git a/modules/seq/base/tests/test_sequence.cc b/modules/seq/base/tests/test_sequence.cc
index 462747784..309939db0 100644
--- a/modules/seq/base/tests/test_sequence.cc
+++ b/modules/seq/base/tests/test_sequence.cc
@@ -106,6 +106,26 @@ BOOST_AUTO_TEST_CASE(seq_string)
BOOST_CHECK_EQUAL(s.GetGaplessString(),"");
}
+BOOST_AUTO_TEST_CASE(seq_normalise)
+{
+ SequenceHandle s=CreateSequence("S1", "abfcdadeaf");
+ BOOST_CHECK_EQUAL(s.GetString(),"abfcdadeaf");
+ s.Normalise();
+ BOOST_CHECK_EQUAL(s.GetString(),"ABFCDADEAF");
+ s=CreateSequence("S1", ".afc..de.f");
+ BOOST_CHECK_EQUAL(s.GetString(),".afc..de.f");
+ s.Normalise();
+ BOOST_CHECK_EQUAL(s.GetString(),"AFCDEF");
+ s=CreateSequence("S1", "-afc--de-f");
+ BOOST_CHECK_EQUAL(s.GetString(),"-afc--de-f");
+ s.Normalise();
+ BOOST_CHECK_EQUAL(s.GetString(),"AFCDEF");
+ s=CreateSequence("S1", ".afc-.de-f");
+ BOOST_CHECK_EQUAL(s.GetString(),".afc-.de-f");
+ s.Normalise();
+ BOOST_CHECK_EQUAL(s.GetString(),"AFCDEF");
+}
+
BOOST_AUTO_TEST_CASE(seq_onelettercode)
{
SequenceHandle s=CreateSequence("S1", "abfcdadeaf");
--
GitLab