RNApolis 0.4.8__tar.gz → 0.4.10__tar.gz

{rnapolis-0.4.8/src/RNApolis.egg-info → rnapolis-0.4.10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: RNApolis
-Version: 0.4.8
+Version: 0.4.10
 Summary: A Python library containing RNA-related bioinformatics functions and classes
 Home-page: https://github.com/tzok/rnapolis-py
 Author: Tomasz Zok

{rnapolis-0.4.8 → rnapolis-0.4.10}/setup.py

@@ -5,7 +5,7 @@ with open("README.md") as f:
 
 setup(
     name="RNApolis",
-    version="0.4.8",
+    version="0.4.10",
     packages=["rnapolis"],
     package_dir={"": "src"},
     author="Tomasz Zok",

{rnapolis-0.4.8 → rnapolis-0.4.10/src/RNApolis.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: RNApolis
-Version: 0.4.8
+Version: 0.4.10
 Summary: A Python library containing RNA-related bioinformatics functions and classes
 Home-page: https://github.com/tzok/rnapolis-py
 Author: Tomasz Zok

{rnapolis-0.4.8 → rnapolis-0.4.10}/src/rnapolis/annotator.py

@@ -9,6 +9,7 @@ from typing import Dict, List, Optional, Set, Tuple
 
 import numpy
 import numpy.typing
+import orjson
 from ordered_set import OrderedSet
 from rnapolis.common import (
     BR,

{rnapolis-0.4.8 → rnapolis-0.4.10}/src/rnapolis/common.py

@@ -940,6 +940,27 @@ class BpSeq:
             solutions.add(self.__make_dot_bracket(regions, orders))
         return list(solutions)
 
+    def without_pseudoknots(self):
+        return BpSeq.from_dotbracket(self.dot_bracket.without_pseudoknots())
+
+    def without_isolated(self):
+        stems, _, _, _ = self.elements
+        to_unpair = []
+
+        for stem in stems:
+            if stem.strand5p.first == stem.strand5p.last:
+                to_unpair.append(stem.strand5p.first - 1)
+                to_unpair.append(stem.strand3p.first - 1)
+
+        if not to_unpair:
+            return self
+
+        entries = self.entries.copy()
+        for i in to_unpair:
+            entries[i].pair = 0
+
+        return BpSeq(entries)
+
 
 @dataclass
 class DotBracket:
@@ -990,6 +1011,10 @@ class DotBracket:
     def __hash__(self) -> int:
         return hash((self.sequence, self.structure))
 
+    def without_pseudoknots(self):
+        structure = re.sub(r"[\[\]\{\}\<\>A-Za-z]", ".", self.structure)
+        return DotBracket(self.sequence, structure)
+
 
 @dataclass
 class MultiStrandDotBracket(DotBracket):

{rnapolis-0.4.8 → rnapolis-0.4.10}/src/rnapolis/motif_extractor.py

@@ -9,6 +9,12 @@ def main():
     parser = argparse.ArgumentParser()
     parser.add_argument("--dbn", help="path to DotBracket file")
     parser.add_argument("--bpseq", help="path to BpSeq file")
+    parser.add_argument(
+        "--remove-pseudoknots", action="store_true", help="remove pseudoknots"
+    )
+    parser.add_argument(
+        "--remove-isolated", action="store_true", help="remove isolated base pairs"
+    )
     args = parser.parse_args()
 
     if args.dbn:
@@ -19,6 +25,12 @@ def main():
         parser.print_help()
         return
 
+    if args.remove_isolated:
+        bpseq = bpseq.without_isolated()
+
+    if args.remove_pseudoknots:
+        bpseq = bpseq.without_pseudoknots()
+
     print(f"Full dot-bracket:\n{bpseq.dot_bracket}")
     stems, single_strands, hairpins, loops = bpseq.elements
 

{rnapolis-0.4.8 → rnapolis-0.4.10}/tests/test_common.py

@@ -199,3 +199,37 @@ def test_high_level_pseudoknot():
         dot_bracket.structure
         == "([{<" + string.ascii_uppercase + ")]}>" + string.ascii_lowercase
     )
+
+
+def test_bpseq_removal_options():
+    sequence = (
+        "GCGGAUUUAGCUCAGUUGGGAGAGCGCCAGACUGAAGAUCUGGAGGUCCUGUGUUCCAUCCACAGAAUUCGCACCA"
+    )
+    structure = (
+        "(((((((..((((....[[..)))).((((..(...)..)))).....(((((..]]...))))))))))))...."
+    )
+
+    bpseq = BpSeq.from_dotbracket(DotBracket(sequence, structure))
+    assert bpseq.dot_bracket.sequence == sequence
+    assert bpseq.dot_bracket.structure == structure
+
+    bpseq_without_isolated = bpseq.without_isolated()
+    assert bpseq_without_isolated.dot_bracket.sequence == sequence
+    assert (
+        bpseq_without_isolated.dot_bracket.structure
+        == "(((((((..((((....[[..)))).((((.........)))).....(((((..]]...))))))))))))...."
+    )
+
+    bpseq_without_pseudoknots = bpseq.without_pseudoknots()
+    assert bpseq_without_pseudoknots.dot_bracket.sequence == sequence
+    assert (
+        bpseq_without_pseudoknots.dot_bracket.structure
+        == "(((((((..((((........)))).((((..(...)..)))).....(((((.......))))))))))))...."
+    )
+
+    bpseq_without_both = bpseq.without_isolated().without_pseudoknots()
+    assert bpseq_without_both.dot_bracket.sequence == sequence
+    assert (
+        bpseq_without_both.dot_bracket.structure
+        == "(((((((..((((........)))).((((.........)))).....(((((.......))))))))))))...."
+    )