ptm sphere

backend/main/views_helper.py

@@ -5,6 +5,10 @@
 import pandas as pd
 
 from backend.protzilla.constants.paths import SETTINGS_PATH
+from backend.protzilla.data_analysis.amino_acid_spheres import (
+    calculate_amino_acid_spheres,
+    calculate_ptm_spheres,
+)
 from backend.protzilla.disk_operator import YamlOperator
 from backend.protzilla.steps import Step
 from backend.protzilla.step_manager import StepManager
@@ -211,6 +215,8 @@ def create_visualization(
     cif_df: pd.DataFrame,
     structure_entry_id: str,
     crosslinking_df: pd.DataFrame | None = None,
+    include_ptm_spheres: bool = False,
+    ignored_neighbors: int = 0,
 ) -> dict:
     """
     Create visualization data, by packaging a mmCIF string (converted from a CIF DataFrame) with its structure entry ID.
@@ -233,6 +239,14 @@ def create_visualization(
 
     result = {"structureEntryId": structure_entry_id, "cifString": cif_string}
 
+    if include_ptm_spheres:
+        result["trimeshMeshes"] = calculate_amino_acid_spheres(
+            cif_df,
+            only_intersecting_ptms=True,
+            ignored_neighbors=int(ignored_neighbors),
+        )
+        result["trimeshMeshes"].extend(calculate_ptm_spheres(cif_df))
+
     if crosslinking_df is not None:
         result["crosslinks"] = extract_relevant_crosslink_information(crosslinking_df)
 

backend/protzilla/all_steps.py

@@ -82,6 +82,7 @@
     data_analysis.PTMDetailsVisualization,
     data_analysis.CrosslinkingValidationWithAngstromDeviation,
     data_analysis.CrosslinkingValidationWithAngstromDeviationForMultimer,
+    data_analysis.PtmValidation,
     data_preprocessing.ImputationByMinPerSample,
     data_integration.EnrichmentAnalysisGOAnalysisWithString,
     data_integration.EnrichmentAnalysisGOAnalysisWithEnrichr,

backend/protzilla/constants/cif_columns.py

@@ -30,6 +30,7 @@ class ATOM_SITE_COLUMNS(StrEnum):
     AUTH_ASYM_ID = f"{ATOM_SITE_PREFIX}auth_asym_id"
     AUTH_ATOM_ID = f"{ATOM_SITE_PREFIX}auth_atom_id"
     PDBX_PDB_MODEL_NUM = f"{ATOM_SITE_PREFIX}pdbx_PDB_model_num"
+    PDBX_SIFTS_XREF_DB_ACC = f"{ATOM_SITE_PREFIX}pdbx_sifts_xref_db_acc"
 
 
 ATOM_SITE_LABEL_COMP_ID = ATOM_SITE_COLUMNS.LABEL_COMP_ID

backend/protzilla/constants/van_der_waals.py

@@ -0,0 +1,123 @@
+vdw_radii = {
+    """
+    Van der Waals radius in Ångström
+    """
+    "H": 1.20,
+    "He": 1.43,
+    "Li": 2.12,
+    "Be": 1.98,
+    "B": 1.91,
+    "C": 1.77,
+    "N": 1.66,
+    "O": 1.50,
+    "F": 1.46,
+    "Ne": 1.58,
+    "Na": 2.50,
+    "Mg": 2.51,
+    "Al": 2.25,
+    "Si": 2.19,
+    "P": 1.90,
+    "S": 1.89,
+    "Cl": 1.82,
+    "Ar": 1.83,
+    "K": 2.73,
+    "Ca": 2.62,
+    "Sc": 2.58,
+    "Ti": 2.46,
+    "V": 2.42,
+    "Cr": 2.45,
+    "Mn": 2.45,
+    "Fe": 2.44,
+    "Co": 2.40,
+    "Ni": 2.40,
+    "Cu": 2.38,
+    "Zn": 2.39,
+    "Ga": 2.32,
+    "Ge": 2.29,
+    "As": 1.88,
+    "Se": 1.82,
+    "Br": 1.86,
+    "Kr": 2.25,
+    "Rb": 3.21,
+    "Sr": 2.84,
+    "Y": 2.75,
+    "Zr": 2.52,
+    "Nb": 2.56,
+    "Mo": 2.45,
+    "Tc": 2.44,
+    "Ru": 2.46,
+    "Rh": 2.44,
+    "Pd": 2.15,
+    "Ag": 2.53,
+    "Cd": 2.49,
+    "In": 2.43,
+    "Sn": 2.42,
+    "Sb": 2.47,
+    "Te": 1.99,
+    "I": 2.04,
+    "Xe": 2.06,
+    "Cs": 3.48,
+    "Ba": 3.03,
+    "La": 2.98,
+    "Ce": 2.88,
+    "Pr": 2.92,
+    "Nd": 2.95,
+    "Pm": None,
+    "Sm": 2.90,
+    "Eu": 2.87,
+    "Gd": 2.83,
+    "Tb": 2.79,
+    "Dy": 2.87,
+    "Ho": 2.81,
+    "Er": 2.83,
+    "Tm": 2.79,
+    "Yb": 2.80,
+    "Lu": 2.74,
+    "Hf": 2.63,
+    "Ta": 2.53,
+    "W": 2.57,
+    "Re": 2.49,
+    "Os": 2.48,
+    "Ir": 2.41,
+    "Pt": 2.29,
+    "Au": 2.32,
+    "Hg": 2.45,
+    "Tl": 2.47,
+    "Pb": 2.60,
+    "Bi": 2.54,
+    "Po": None,
+    "At": None,
+    "Rn": None,
+    "Fr": None,
+    "Ra": None,
+    "Ac": 2.80,
+    "Th": 2.93,
+    "Pa": 2.88,
+    "U": 2.71,
+    "Np": 2.82,
+    "Pu": 2.81,
+    "Am": 2.83,
+    "Cm": 3.05,
+    "Bk": 3.40,
+    "Cf": 3.05,
+    "Es": 2.70,
+    "Fm": None,
+    "Md": None,
+    "No": None,
+    "Lr": None,
+    "Rf": None,
+    "Db": None,
+    "Sg": None,
+    "Bh": None,
+    "Hs": None,
+    "Mt": None,
+    "Ds": None,
+    "Rg": None,
+    "Cn": None,
+    "Nh": None,
+    "Fl": None,
+    "Mc": None,
+    "Lv": None,
+    "Ts": None,
+    "Og": None,
+}

backend/protzilla/data_analysis/amino_acid_spheres.py

@@ -0,0 +1,194 @@
+from __future__ import annotations
+
+import numpy as np
+import pandas as pd
+import trimesh
+
+from backend.protzilla.constants.cif_columns import (
+    ATOM_SITE_COLUMNS,
+    CHEM_COMP_COLUMNS,
+)
+from backend.protzilla.data_analysis.geometry_operations import (
+    calculate_center_point,
+    extract_points_from_cif,
+    find_farthest_point_vdw,
+    find_intersecting_spheres,
+    mesh_to_polyhedron,
+    resolve_chain_column,
+)
+from backend.protzilla.utilities.ptm_helper import (
+    get_all_ptm_atoms_with_coordinates,
+    get_center_points_and_radius_for_each_ptm,
+)
+
+
+def _calculate_residue_spheres(
+    cif_df: pd.DataFrame, chain_id: str | None = None
+) -> list[dict]:
+    residue_df = cif_df
+    chain_column = resolve_chain_column(cif_df)
+    if chain_id is not None and chain_column is not None:
+        residue_df = cif_df[cif_df[chain_column] == chain_id]
+
+    residue_positions = (
+        pd.to_numeric(residue_df[ATOM_SITE_COLUMNS.LABEL_SEQ_ID], errors="coerce")
+        .dropna()
+        .astype(int)
+        .drop_duplicates()
+        .sort_values()
+    )
+
+    spheres = []
+    for residue_position in residue_positions:
+        residue_points, residue_elements = extract_points_from_cif(
+            cif_df,
+            residue_range=(residue_position, residue_position),
+            chain_id=chain_id,
+        )
+        center = calculate_center_point(residue_points)
+        _, radius = find_farthest_point_vdw(residue_points, residue_elements, center)
+
+        spheres.append(
+            {
+                "chain": chain_id,
+                "position": residue_position,
+                "center": center.tolist(),
+                "radius": float(radius),
+            }
+        )
+
+    return spheres
+
+
+def calculate_amino_acid_spheres(
+    cif_df: pd.DataFrame,
+    chain_id: str | None = None,
+    color: int = 0xFF8C00,
+    alpha: float = 0.25,
+    subdivisions: int = 1,
+    only_intersecting_ptms: bool = False,
+    ignored_neighbors: int = 0,
+) -> list[dict]:
+    if only_intersecting_ptms:
+        residue_spheres = []
+        seen_residues = set()
+        for collision in find_ptm_amino_acid_sphere_collisions(
+            cif_df, ignored_neighbors=ignored_neighbors, chain_id=chain_id
+        ):
+            for residue_sphere in collision["collisions"]:
+                residue_key = (residue_sphere["chain"], residue_sphere["position"])
+                if residue_key not in seen_residues:
+                    seen_residues.add(residue_key)
+                    residue_spheres.append(residue_sphere)
+    else:
+        residue_spheres = _calculate_residue_spheres(cif_df, chain_id)
+
+    spheres = []
+    for residue_sphere in residue_spheres:
+        sphere = trimesh.creation.icosphere(
+            subdivisions=subdivisions, radius=residue_sphere["radius"]
+        )
+        sphere.apply_translation(residue_sphere["center"])
+
+        spheres.append(
+            {
+                "label": f"Residue {residue_sphere['position']} sphere",
+                "mesh": mesh_to_polyhedron(sphere),
+                "color": color,
+                "alpha": alpha,
+            }
+        )
+
+    return spheres
+
+
+def calculate_ptm_spheres(
+    cif_df: pd.DataFrame,
+    color: int = 0x00A6A6,
+    alpha: float = 0.35,
+    subdivisions: int = 1,
+) -> list[dict]:
+    if CHEM_COMP_COLUMNS.MON_NSTD_FLAG not in cif_df.columns:
+        return []
+
+    ptms = get_center_points_and_radius_for_each_ptm(
+        get_all_ptm_atoms_with_coordinates(cif_df)
+    )
+
+    spheres = []
+    for ptm in ptms:
+        center = np.array(ptm["center_point"], dtype=float)
+        radius = float(ptm["radius"])
+
+        sphere = trimesh.creation.icosphere(subdivisions=subdivisions, radius=radius)
+        sphere.apply_translation(center)
+
+        spheres.append(
+            {
+                "label": f"PTM {ptm['ptm_name']} {ptm['chain']}:{ptm['position']} sphere",
+                "mesh": mesh_to_polyhedron(sphere),
+                "color": color,
+                "alpha": alpha,
+            }
+        )
+
+    return spheres
+
+
+def find_ptm_amino_acid_sphere_collisions(
+    cif_df: pd.DataFrame,
+    ignored_neighbors: int = 0,
+    chain_id: str | None = None,
+) -> list[dict]:
+    if CHEM_COMP_COLUMNS.MON_NSTD_FLAG not in cif_df.columns:
+        return []
+
+    amino_acid_df = cif_df[cif_df[CHEM_COMP_COLUMNS.MON_NSTD_FLAG] == True]
+    chain_column = resolve_chain_column(amino_acid_df)
+    if chain_id is not None:
+        chains = [chain_id]
+    elif chain_column is not None:
+        chains = amino_acid_df[chain_column].dropna().drop_duplicates()
+    else:
+        chains = [None]
+    amino_acid_spheres = [
+        sphere
+        for chain in chains
+        for sphere in _calculate_residue_spheres(amino_acid_df, chain)
+    ]
+
+    ptms = get_center_points_and_radius_for_each_ptm(
+        get_all_ptm_atoms_with_coordinates(cif_df)
+    )
+    if chain_id is not None:
+        ptms = [ptm for ptm in ptms if ptm["chain"] == chain_id]
+
+    collisions = []
+    for ptm in ptms:
+        ptm_sphere = {
+            "ptm_name": ptm["ptm_name"],
+            "chain": ptm["chain"],
+            "position": int(ptm["position"]),
+            "center": ptm["center_point"],
+            "radius": float(ptm["radius"]),
+        }
+        checked_spheres = [
+            sphere
+            for sphere in amino_acid_spheres
+            if not (
+                sphere["chain"] == ptm_sphere["chain"]
+                and 0
+                < abs(sphere["position"] - ptm_sphere["position"])
+                <= ignored_neighbors
+            )
+        ]
+        intersecting_spheres = find_intersecting_spheres(checked_spheres, ptm_sphere)
+
+        collisions.append(
+            {
+                **ptm_sphere,
+                "collisions": intersecting_spheres,
+            }
+        )
+
+    return collisions