RNApolis 0.10.4__py3-none-any.whl → 0.10.5__py3-none-any.whl

rnapolis/adapter.py

@@ -317,10 +317,14 @@ def parse_dssr_output(
             if nt1 is not None and nt2 is not None:
                 stackings.append(Stacking(nt1, nt2, None))
 
-    return BaseInteractions(base_pairs, stackings, [], [], [])
+    return BaseInteractions.from_structure3d(
+        structure3d, base_pairs, stackings, [], [], []
+    )
 
 
-def parse_maxit_output(file_paths: List[str]) -> BaseInteractions:
+def parse_maxit_output(
+    file_paths: List[str], structure3d: Structure3D
+) -> BaseInteractions:
     """
     Parse MAXIT output files and convert to BaseInteractions.
 
@@ -448,10 +452,14 @@ def parse_maxit_output(file_paths: List[str]) -> BaseInteractions:
     except Exception as e:
         logging.warning(f"Error processing MAXIT file {cif_file}: {e}", exc_info=True)
 
-    return BaseInteractions(all_base_pairs, [], [], [], all_other_interactions)
+    return BaseInteractions.from_structure3d(
+        structure3d, all_base_pairs, [], [], [], all_other_interactions
+    )
 
 
-def parse_bpnet_output(file_paths: List[str]) -> BaseInteractions:
+def parse_bpnet_output(
+    file_paths: List[str], structure3d: Structure3D
+) -> BaseInteractions:
     """
     Parse BPNet output files and convert to BaseInteractions.
 
@@ -649,7 +657,8 @@ def parse_bpnet_output(file_paths: List[str]) -> BaseInteractions:
                 f"Error processing BPNet rob file {rob_file}: {e}", exc_info=True
             )
 
-    return BaseInteractions(
+    return BaseInteractions.from_structure3d(
+        structure3d,
         base_pairs,
         stackings,
         base_ribose_interactions,
@@ -986,7 +995,8 @@ def parse_rnaview_output(
     except Exception as e:
         logging.warning(f"Error processing RNAView file {out_file}: {e}", exc_info=True)
 
-    return BaseInteractions(
+    return BaseInteractions.from_structure3d(
+        structure3d,
         base_pairs,
         stackings,
         base_ribose_interactions,
@@ -1142,7 +1152,9 @@ def parse_barnaba_output(
                         f"Unknown barnaba stacking topology: {interaction_str}"
                     )
 
-    return BaseInteractions(base_pairs, stackings, [], [], other_interactions)
+    return BaseInteractions.from_structure3d(
+        structure3d, base_pairs, stackings, [], [], other_interactions
+    )
 
 
 def parse_external_output(
@@ -1160,13 +1172,13 @@ def parse_external_output(
         BaseInteractions object containing the interactions found by the external tool
     """
     if tool == ExternalTool.FR3D:
-        return parse_fr3d_output(file_paths)
+        return parse_fr3d_output(file_paths, structure3d)
     elif tool == ExternalTool.DSSR:
         return parse_dssr_output(file_paths, structure3d)
     elif tool == ExternalTool.MAXIT:
-        return parse_maxit_output(file_paths)
+        return parse_maxit_output(file_paths, structure3d)
     elif tool == ExternalTool.BPNET:
-        return parse_bpnet_output(file_paths)
+        return parse_bpnet_output(file_paths, structure3d)
     elif tool == ExternalTool.RNAVIEW:
         return parse_rnaview_output(file_paths, structure3d)
     elif tool == ExternalTool.BARNABA:
@@ -1175,7 +1187,9 @@ def parse_external_output(
         raise ValueError(f"Unsupported external tool: {tool}")
 
 
-def parse_fr3d_output(file_paths: List[str]) -> BaseInteractions:
+def parse_fr3d_output(
+    file_paths: List[str], structure3d: Structure3D
+) -> BaseInteractions:
     """
     Parse FR3D output files and convert to BaseInteractions.
 
@@ -1208,7 +1222,8 @@ def parse_fr3d_output(file_paths: List[str]) -> BaseInteractions:
                 _process_interaction_line(line, interactions_data)
 
     # Return a BaseInteractions object with all the processed interactions
-    return BaseInteractions(
+    return BaseInteractions.from_structure3d(
+        structure3d,
         interactions_data["base_pairs"],
         interactions_data["stackings"],
         interactions_data["base_ribose_interactions"],

rnapolis/annotator.py

@@ -85,15 +85,6 @@ def detect_cis_trans(residue_i: Residue3D, residue_j: Residue3D) -> Optional[str
     return "c" if -90.0 < torsion < 90.0 else "t"
 
 
-def detect_saenger(
-    residue_i: Residue3D, residue_j: Residue3D, lw: LeontisWesthof
-) -> Optional[Saenger]:
-    key = (f"{residue_i.one_letter_name}{residue_j.one_letter_name}", lw.value)
-    if key in Saenger.table():
-        return Saenger[Saenger.table()[key]]
-    return None
-
-
 def detect_bph_br_classification(
     donor_residue: Residue3D, donor: Atom, acceptor: Atom
 ) -> Optional[int]:
@@ -367,7 +358,9 @@ def find_pairs(
                 Residue(residue_i.label, residue_i.auth),
                 Residue(residue_j.label, residue_j.auth),
                 lw,
-                detect_saenger(residue_i, residue_j, lw),
+                Saenger.from_leontis_westhof(
+                    residue_i.one_letter_name, residue_j.one_letter_name, lw
+                ),
             )
         )
 
@@ -483,7 +476,9 @@ def extract_base_interactions(
 ) -> BaseInteractions:
     base_pairs, base_phosphate, base_ribose = find_pairs(tertiary_structure, model)
     stackings = find_stackings(tertiary_structure, model)
-    return BaseInteractions(base_pairs, stackings, base_ribose, base_phosphate, [])
+    return BaseInteractions.from_structure3d(
+        tertiary_structure, base_pairs, stackings, base_ribose, base_phosphate, []
+    )
 
 
 def generate_pymol_script(mapping: Mapping2D3D, stems: List[Stem]) -> str:
@@ -688,91 +683,6 @@ def add_common_output_arguments(parser: argparse.ArgumentParser):
     )
 
 
-def unify_structure_data(structure2d: Structure2D, mapping: Mapping2D3D) -> Structure2D:
-    """
-    Unify structure data by:
-    1. Adding missing Saenger classifications to base pairs
-    2. Filling in empty residue labels from Structure3D
-    """
-    # Create a mapping from residue to residue3d for label filling
-    residue_to_residue3d = {}
-    for residue3d in mapping.structure3d.residues:
-        residue_key = Residue(residue3d.label, residue3d.auth)
-        residue_to_residue3d[residue_key] = residue3d
-
-    def fill_residue_label(residue: Residue) -> Residue:
-        """Fill empty label from Structure3D if available."""
-        if residue.label is not None:
-            return residue
-
-        # Try to find matching residue3d by auth
-        for residue3d in mapping.structure3d.residues:
-            if residue.auth == residue3d.auth:
-                return Residue(residue3d.label, residue.auth)
-
-        return residue
-
-    # Process base pairs
-    unified_base_pairs = []
-    for base_pair in structure2d.base_pairs:
-        # Fill in missing labels
-        nt1 = fill_residue_label(base_pair.nt1)
-        nt2 = fill_residue_label(base_pair.nt2)
-
-        # Detect missing Saenger classification
-        saenger = base_pair.saenger
-        if saenger is None:
-            # Find corresponding 3D residues for Saenger detection
-            residue3d_1 = residue_to_residue3d.get(Residue(nt1.label, nt1.auth))
-            residue3d_2 = residue_to_residue3d.get(Residue(nt2.label, nt2.auth))
-
-            if residue3d_1 is not None and residue3d_2 is not None:
-                saenger = detect_saenger(residue3d_1, residue3d_2, base_pair.lw)
-
-        unified_base_pairs.append(BasePair(nt1, nt2, base_pair.lw, saenger))
-
-    # Process other interaction types (fill labels only)
-    unified_stackings = []
-    for stacking in structure2d.stackings:
-        nt1 = fill_residue_label(stacking.nt1)
-        nt2 = fill_residue_label(stacking.nt2)
-        unified_stackings.append(Stacking(nt1, nt2, stacking.topology))
-
-    unified_base_ribose = []
-    for base_ribose in structure2d.base_ribose_interactions:
-        nt1 = fill_residue_label(base_ribose.nt1)
-        nt2 = fill_residue_label(base_ribose.nt2)
-        unified_base_ribose.append(BaseRibose(nt1, nt2, base_ribose.br))
-
-    unified_base_phosphate = []
-    for base_phosphate in structure2d.base_phosphate_interactions:
-        nt1 = fill_residue_label(base_phosphate.nt1)
-        nt2 = fill_residue_label(base_phosphate.nt2)
-        unified_base_phosphate.append(BasePhosphate(nt1, nt2, base_phosphate.bph))
-
-    unified_other = []
-    for other in structure2d.other_interactions:
-        nt1 = fill_residue_label(other.nt1)
-        nt2 = fill_residue_label(other.nt2)
-        unified_other.append(OtherInteraction(nt1, nt2))
-
-    # Create new Structure2D with unified data
-    unified_base_interactions = BaseInteractions(
-        unified_base_pairs,
-        unified_stackings,
-        unified_base_ribose,
-        unified_base_phosphate,
-        unified_other,
-    )
-
-    # Recreate Structure2D with unified interactions
-    unified_structure2d, _ = mapping.structure3d.extract_secondary_structure(
-        unified_base_interactions, False
-    )
-
-    return unified_structure2d
-
-
 def handle_output_arguments(
     args: argparse.Namespace,
     structure2d: Structure2D,
@@ -780,34 +690,31 @@ def handle_output_arguments(
     input_filename: str,
 ):
     """Handles writing output based on provided arguments."""
-    # Unify the structure data before processing outputs
-    unified_structure2d = unify_structure_data(structure2d, mapping)
-
     input_basename = os.path.basename(input_filename)
     if args.csv:
-        write_csv(args.csv, unified_structure2d)
+        write_csv(args.csv, structure2d)
 
     if args.json:
-        write_json(args.json, unified_structure2d)
+        write_json(args.json, structure2d)
 
     if args.bpseq:
-        write_bpseq(args.bpseq, unified_structure2d.bpseq)
+        write_bpseq(args.bpseq, structure2d.bpseq)
 
     if args.extended:
-        print(unified_structure2d.extended_dot_bracket)
+        print(structure2d.extended_dot_bracket)
     else:
-        print(unified_structure2d.dot_bracket)
+        print(structure2d.dot_bracket)
 
     if args.dot:
-        print(BpSeq.from_string(unified_structure2d.bpseq).graphviz)
+        print(BpSeq.from_string(structure2d.bpseq).graphviz)
 
     if args.pml:
-        pml_script = generate_pymol_script(mapping, unified_structure2d.stems)
+        pml_script = generate_pymol_script(mapping, structure2d.stems)
         with open(args.pml, "w") as f:
             f.write(pml_script)
 
     if args.inter_stem_csv:
-        if unified_structure2d.inter_stem_parameters:
+        if structure2d.inter_stem_parameters:
             # Convert list of dataclasses to list of dicts
             params_list = [
                 {
@@ -820,7 +727,7 @@ def handle_output_arguments(
                     "min_endpoint_distance_pdf": p.min_endpoint_distance_pdf,
                     "coaxial_probability": p.coaxial_probability,
                 }
-                for p in unified_structure2d.interStemParameters
+                for p in structure2d.interStemParameters
             ]
             df = pd.DataFrame(params_list)
             df["input_basename"] = input_basename
@@ -838,9 +745,9 @@ def handle_output_arguments(
             # pd.DataFrame(columns=['input_basename', 'stem1_idx', ...]).to_csv(args.inter_stem_csv, index=False)
 
     if args.stems_csv:
-        if unified_structure2d.stems:
+        if structure2d.stems:
             stems_data = []
-            for i, stem in enumerate(unified_structure2d.stems):
+            for i, stem in enumerate(structure2d.stems):
                 try:
                     res5p_first = mapping.bpseq_index_to_residue_map.get(
                         stem.strand5p.first

rnapolis/common.py

@@ -5,7 +5,7 @@ import re
 import string
 from collections import defaultdict
 from collections.abc import Sequence
-from dataclasses import dataclass
+from dataclasses import InitVar, dataclass
 from enum import Enum
 from functools import cache, cached_property, total_ordering
 from typing import Dict, List, Optional, Tuple
@@ -152,6 +152,18 @@ class Saenger(Enum):
             ("TG", "cWW"): "XXVIII",
         }
 
+    @classmethod
+    def from_leontis_westhof(
+        cls,
+        residue_i_one_letter_name: str,
+        residue_j_one_letter_name: str,
+        lw: LeontisWesthof,
+    ) -> Optional["Saenger"]:
+        key = (f"{residue_i_one_letter_name}{residue_j_one_letter_name}", lw.value)
+        if key in Saenger.table():
+            return Saenger[Saenger.table()[key]]
+        return None
+
     @property
     def is_canonical(self) -> bool:
         return self == Saenger.XIX or self == Saenger.XX or self == Saenger.XXVIII
@@ -1062,6 +1074,91 @@ class BaseInteractions:
     base_phosphate_interactions: List[BasePhosphate]
     other_interactions: List[OtherInteraction]
 
+    @classmethod
+    def from_structure3d(
+        cls,
+        structure3d: "Structure3D",
+        base_pairs: List[BasePair],
+        stackings: List[Stacking],
+        base_ribose_interactions: List[BaseRibose],
+        base_phosphate_interactions: List[BasePhosphate],
+        other_interactions: List[OtherInteraction],
+    ) -> "BaseInteractions":
+        auth2residue3d = {}
+        auth2label = {}
+        label2auth = {}
+
+        for residue3d in structure3d.residues:
+            auth2residue3d[residue3d.auth] = residue3d
+            auth2label[residue3d.auth] = residue3d.label
+            label2auth[residue3d.label] = residue3d.auth
+
+        def unify_nt(nt: Residue) -> Residue:
+            if nt.auth is not None and nt.label is not None:
+                return nt
+            if nt.auth is not None:
+                return Residue(label=auth2label.get(nt.auth, None), auth=nt.auth)
+            if nt.label is not None:
+                return Residue(label=nt.label, auth=label2auth.get(nt.label, None))
+            return nt
+
+        base_pairs_new = []
+        for base_pair in base_pairs:
+            nt1 = unify_nt(base_pair.nt1)
+            nt2 = unify_nt(base_pair.nt2)
+            saenger = base_pair.saenger or Saenger.from_leontis_westhof(
+                auth2residue3d[nt1.auth].one_letter_name,
+                auth2residue3d[nt2.auth].one_letter_name,
+                base_pair.lw,
+            )
+            if (
+                nt1 != base_pair.nt1
+                or nt2 != base_pair.nt2
+                or saenger != base_pair.saenger
+            ):
+                base_pair = BasePair(nt1=nt1, nt2=nt2, lw=base_pair.lw, saenger=saenger)
+            base_pairs_new.append(base_pair)
+
+        stackings_new = []
+        for stacking in stackings:
+            nt1 = unify_nt(stacking.nt1)
+            nt2 = unify_nt(stacking.nt2)
+            if nt1 != stacking.nt1 or nt2 != stacking.nt2:
+                stacking = Stacking(nt1=nt1, nt2=nt2, topology=stacking.topology)
+            stackings_new.append(stacking)
+
+        base_ribose_interactions_new = []
+        for base_ribose in base_ribose_interactions:
+            nt1 = unify_nt(base_ribose.nt1)
+            nt2 = unify_nt(base_ribose.nt2)
+            if nt1 != base_ribose.nt1 or nt2 != base_ribose.nt2:
+                base_ribose = BaseRibose(nt1=nt1, nt2=nt2, br=base_ribose.br)
+            base_ribose_interactions_new.append(base_ribose)
+
+        base_phosphate_interactions_new = []
+        for base_phosphate in base_phosphate_interactions:
+            nt1 = unify_nt(base_phosphate.nt1)
+            nt2 = unify_nt(base_phosphate.nt2)
+            if nt1 != base_phosphate.nt1 or nt2 != base_phosphate.nt2:
+                base_phosphate = BasePhosphate(nt1=nt1, nt2=nt2, bph=base_phosphate.bph)
+            base_phosphate_interactions_new.append(base_phosphate)
+
+        other_interactions_new = []
+        for other_interaction in other_interactions:
+            nt1 = unify_nt(other_interaction.nt1)
+            nt2 = unify_nt(other_interaction.nt2)
+            if nt1 != other_interaction.nt1 or nt2 != other_interaction.nt2:
+                other_interaction = OtherInteraction(nt1=nt1, nt2=nt2)
+            other_interactions_new.append(other_interaction)
+
+        return cls(
+            base_pairs=base_pairs_new,
+            stackings=stackings_new,
+            base_ribose_interactions=base_ribose_interactions_new,
+            base_phosphate_interactions=base_phosphate_interactions_new,
+            other_interactions=other_interactions_new,
+        )
+
 
 @dataclass(frozen=True, order=True)
 class InterStemParameters:

{rnapolis-0.10.4.dist-info → rnapolis-0.10.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: RNApolis
-Version: 0.10.4
+Version: 0.10.5
 Summary: A Python library containing RNA-related bioinformatics functions and classes
 Home-page: https://github.com/tzok/rnapolis-py
 Author: Tomasz Zok

{rnapolis-0.10.4.dist-info → rnapolis-0.10.5.dist-info}/RECORD

@@ -1,8 +1,8 @@
-rnapolis/adapter.py,sha256=6b4m1EzD1BA3D7GgXaV4iYKOnmzgz0AKwXHc2svQj3w,48132
+rnapolis/adapter.py,sha256=0Awt6owSeGcWq8kgLtkDv19suhrmcJgQip_Hi4Y3tK4,48513
 rnapolis/aligner.py,sha256=o7rQyjAZ3n4VXcnSPY3HVB8nLNRkVbl552O3NVh0mfg,3429
-rnapolis/annotator.py,sha256=OkqFVuxOtb-mySmw3bc5NF9ETu4BWq4ImtBecWJikrY,33899
+rnapolis/annotator.py,sha256=HA2hfEUXdmBElObqRlASAB1FgkysjiHgwMTjEhsDiDE,30277
 rnapolis/clashfinder.py,sha256=AC9_tIx7QIk57sELq_aKfU1u3UMOXbgcccQeGHhMR6c,8517
-rnapolis/common.py,sha256=HTe-RSZa_9hEIi-j4-1afxdqt7zAD-BpZ7JxRZGX170,32390
+rnapolis/common.py,sha256=hamlW892ZF5A0dSWsl7cOCZqOpbVQMgXjVPYDFzk3pE,36347
 rnapolis/component_A.csv,sha256=koirS-AwUZwoYGItT8yn3wS6Idvmh2FANfTQcOS_xh8,2897
 rnapolis/component_C.csv,sha256=NtvsAu_YrUgTjzZm3j4poW4IZ99x3dPARB09XVIiMCc,2803
 rnapolis/component_G.csv,sha256=Z5wl8OnHRyx4XhTyBiWgRZiEvmZXhoxtVRH8bn6Vxf0,2898
@@ -22,9 +22,9 @@ rnapolis/tertiary_v2.py,sha256=SgijTv0bPqMJwsMqyQk0O8QAnS2Ozk45vk8igxt9hRs,38001
 rnapolis/transformer.py,sha256=aC0nBmHHJf5TyLvBIV57Jj3tlwpvHbPo347opfAOlQA,3844
 rnapolis/unifier.py,sha256=2ge7IB9FdRgzSAiVD39U_ciwtdDJ2fGzf8mUIudbrqY,5820
 rnapolis/util.py,sha256=IdquFO3PV1_KDqodjupzm0Rqvgy0CeSzxGHaGEHYXVU,543
-rnapolis-0.10.4.dist-info/licenses/LICENSE,sha256=ZGRu12MzCgbYA-Lt8MyBlmjvPZh7xfiD5u5wBx0enq4,1066
-rnapolis-0.10.4.dist-info/METADATA,sha256=VKy39unD-Kyqzg7J7ADgFlseV3FftWCyBtjn-vnYbEU,54611
-rnapolis-0.10.4.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-rnapolis-0.10.4.dist-info/entry_points.txt,sha256=MZMWnYBUYnis-zWDmFfuA5yXtU3W5YdQrm5HA5LrkeM,474
-rnapolis-0.10.4.dist-info/top_level.txt,sha256=LcO18koxZcWoJ21KDRRRo_tyIbmXL5z61dPitZpy8yc,9
-rnapolis-0.10.4.dist-info/RECORD,,
+rnapolis-0.10.5.dist-info/licenses/LICENSE,sha256=ZGRu12MzCgbYA-Lt8MyBlmjvPZh7xfiD5u5wBx0enq4,1066
+rnapolis-0.10.5.dist-info/METADATA,sha256=LbY2r44uwiIoVASkHv1_TAwwKIOCi8Az33PDuTs7gdg,54611
+rnapolis-0.10.5.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+rnapolis-0.10.5.dist-info/entry_points.txt,sha256=MZMWnYBUYnis-zWDmFfuA5yXtU3W5YdQrm5HA5LrkeM,474
+rnapolis-0.10.5.dist-info/top_level.txt,sha256=LcO18koxZcWoJ21KDRRRo_tyIbmXL5z61dPitZpy8yc,9
+rnapolis-0.10.5.dist-info/RECORD,,