From 64924415d4258b0118e542f95795f6eba1b0c8a7 Mon Sep 17 00:00:00 2001
From: Manuel Lera Ramirez <manulera14@gmail.com>
Date: Fri, 10 Nov 2023 17:28:51 +0000
Subject: [PATCH] closes #147

---
 src/pydna/dseq.py         | 71 ++++++++++++++++++++++++++++++++++++++-
 tests/test_module_dseq.py | 17 ++++++++++
 2 files changed, 87 insertions(+), 1 deletion(-)

diff --git a/src/pydna/dseq.py b/src/pydna/dseq.py
index 88df4491..e3e994aa 100644
--- a/src/pydna/dseq.py
+++ b/src/pydna/dseq.py
@@ -74,7 +74,8 @@ class Dseq(_Seq):
     The most common usage is probably to create a Dseq object as a
     part of a Dseqrecord object (see :class:`pydna.dseqrecord.Dseqrecord`).
 
-    There are three ways of creating a Dseq object directly:
+    There are three ways of creating a Dseq object directly listed below, but you can also
+    use the function Dseq.from_full_sequence_and_overhangs() to create a Dseq:
 
     Only one argument (string):
 
@@ -428,6 +429,70 @@ def from_representation(cls, dsdna: str, *args, **kwargs):
         obj._data = _rc(cb[-max(0, ovhg) or len(cb) :]) + wb + _rc(cb[: max(0, len(cb) - ovhg - len(wb))])
         return obj
 
+    @classmethod
+    def from_full_sequence_and_overhangs(cls, full_sequence: str, crick_ovhg: int, watson_ovhg: int):
+        """Create a linear Dseq object from a full sequence and the 3' overhangs of each strand.
+
+        The order of the parameters is like this because the 3' overhang of the crick strand is the one
+        on the left side of the sequence.
+
+
+        Parameters
+        ----------
+        full_sequence: str
+            The full sequence of the Dseq object.
+
+        crick_ovhg: int
+            The overhang of the crick strand in the 3' end. Equivalent to Dseq.ovhg.
+
+        watson_ovhg: int
+            The overhang of the watson strand in the 5' end.
+
+        Returns
+        -------
+        Dseq
+            A Dseq object.
+
+        Examples
+        --------
+        ```
+        >>> Dseq.from_full_sequence_and_overhangs('AAAAAA', crick_ovhg=2, watson_ovhg=2)
+        Dseq(-6)
+          AAAA
+        TTTT
+        >>> Dseq.from_full_sequence_and_overhangs('AAAAAA', crick_ovhg=-2, watson_ovhg=2)
+        Dseq(-6)
+        AAAAAA
+          TT
+        >>> Dseq.from_full_sequence_and_overhangs('AAAAAA', crick_ovhg=2, watson_ovhg=-2)
+        Dseq(-6)
+          AA
+        TTTTTT
+        >>> Dseq.from_full_sequence_and_overhangs('AAAAAA', crick_ovhg=-2, watson_ovhg=-2)
+        Dseq(-6)
+        AAAA
+          TTTT
+        ```
+        """
+        full_sequence_rev = str(Dseq(full_sequence).reverse_complement())
+        watson = full_sequence
+        crick = full_sequence_rev
+
+        # If necessary, we trim the left side
+        if crick_ovhg < 0:
+            crick = crick[:crick_ovhg]
+        elif crick_ovhg > 0:
+            watson = watson[crick_ovhg:]
+
+        # If necessary, we trim the right side
+        if watson_ovhg < 0:
+            watson = watson[:watson_ovhg]
+        elif watson_ovhg > 0:
+            crick = crick[watson_ovhg:]
+
+        return Dseq(watson, crick=crick, ovhg=crick_ovhg)
+
+
     # @property
     # def ovhg(self):
     #     """The ovhg property. This cannot be set directly, but is a
@@ -942,6 +1007,10 @@ def three_prime_end(self):
             type_ = "blunt"
         return type_, sticky
 
+    def watson_ovhg(self):
+        """Returns the overhang of the watson strand at the three prime."""
+        return len(self.watson) - len(self.crick) + self.ovhg
+
     def __add__(self, other):
         """Simulates ligation between two DNA fragments.
 
diff --git a/tests/test_module_dseq.py b/tests/test_module_dseq.py
index 5162e0bb..db8f4858 100644
--- a/tests/test_module_dseq.py
+++ b/tests/test_module_dseq.py
@@ -768,5 +768,22 @@ def test_translate():
     assert str(x.reverse_complement().translate()) == "LFH"
 
 
+def test_from_full_sequence_and_overhangs():
+    from pydna.dseq import Dseq
+    test_cases = [
+        (2, 2, "AAAA", "TTTT"),
+        (-2, 2, "AAAAAA", "TT" ),
+        (2, -2, "AA", "TTTTTT"),
+        (-2, -2, "AAAA", "TTTT"),
+        (0, 0, "AAAAAA", "TTTTTT"),
+    ]
+    for crick_ovhg, watson_ovhg, watson, crick in test_cases:
+        dseq_1 = Dseq.from_full_sequence_and_overhangs("AAAAAA", crick_ovhg=crick_ovhg, watson_ovhg=watson_ovhg)
+        dseq_2 = Dseq(watson, crick, ovhg=crick_ovhg, circular=False)
+
+        assert dseq_1 == dseq_2
+        assert dseq_2.watson_ovhg() == watson_ovhg
+
+
 if __name__ == "__main__":
     pytest.main([__file__, "-vv", "-s"])