nlm_mesh_tree.py

# For an example of use, see https://gist.github.com/pafonta/d33a0d5d849932f8ceab8b711d995497#gistcomment-3965575.

"""Find MeSH terms in the MeSH tree simply (i.e. without using a graph)."""

from __future__ import annotations

import json
from collections.abc import Iterator
from xml.etree.ElementTree import Element  # nosec

from defusedxml import ElementTree


def is_child(mesh: Element, roots: list[str]) -> bool:
    """Check if a MeSH term is a child of the given MeSH terms in the MeSH tree.

    Parameters
    ----------
    mesh
        The MeSH term.
    roots
        The root MeSH terms for which looking for children.

    Returns
    -------
    bool
        True, if the MeSH term is under one of the root MeSH terms or is one of them.
    """
    for x in mesh.iterfind("TreeNumberList/TreeNumber"):
        for y in roots:
            if x.text.startswith(y):
                return True
    return False


def collect(mesh_tree: ElementTree, roots: list[str]) -> Iterator[str]:
    """Collect MeSH terms which are children of the given MeSH terms in the MeSH tree.

    Parameters
    ----------
    mesh_tree
        The MeSH tree.
    roots
        The root MeSH terms for which looking for children.

    Yields
    ------
    str
        The MeSH terms which are under the root MeSH terms, including them.
    """
    for x in mesh_tree.iter("DescriptorRecord"):
        match = is_child(x, roots)
        if match:
            name = x.find("DescriptorName/String").text
            yield " ".join(name.split())

Example of use

Import utility functions:

from nlm_mesh_tree import *

Parse the MeSH tree:

# wget https://nlmpubs.nlm.nih.gov/projects/mesh/MESH_FILES/xmlmesh/desc2021.xml
mesh_tree: ElementTree = ElementTree.parse("desc2021.xml")

Define the MeSH terms to use as roots:

targets = [
    "A08",
    "A11.650",
    "A11.671",
    "E01.370.376",
    "E05.393.332",
    "E05.599.395.642",
    "E05.629",
    "G01.358.500.249.277",
    "G02.111.820.850",
    "G03.493",
    "G04.580",
    "G04.835.850",
    "G05.308",
    "G07.265",
    "G11.561",
    "H01.158.273.180",
    "L01.313.124",
]

Collect the children MeSH terms (roots ones are included too):

mesh_names = set(collect(mesh_tree, targets))

[Optional] Save the collected MeSH terms:

with open("mesh_names.json", "w") as f:
    ordered = sorted(mesh_names)
    json.dump(ordered, f)