mkt.schema.utils

Schema utility helpers: recursive attribute access, UUID generation, and kinase-group adjudication.

Provides rgetattr()/rsetattr() for traversing nested Pydantic models, random_uuid(), return_kinase_gene_set(), and adjudicate_kinase_group().

Module Attributes

TQDM_BAR_FORMAT	Default tqdm bar format with comma-separated thousands in counts.
SET_HOMOLOG_DO_NOT_MERGE	Hand-curated name pairs that group_name_homologs() must never collapse together, overriding the prefix heuristic.

Functions

adjudicate_kinase_group(str_kinase[, bool_lipid])	Adjudicates the kinase group for a given kinase.
group_name_homologs(names[, min_prefix, ...])	Collapse prefix-homologous kinase names into compact labeled groups.
random_uuid()	Generate a random UUID that allows to set a seed.
return_kinase_gene_set([dict_kinase])	Return the set of kinase HGNC gene symbols for gene-level membership tests.
rgetattr(obj, attr, *args)	Get attribute from object recursively.
rsetattr(obj, attr, val)	Set attribute from object recursively.
split_domain_suffix(name)	Split a trailing multi-domain index suffix off a kinase name.

mkt.schema.utils.SET_HOMOLOG_DO_NOT_MERGE = frozenset({frozenset({'BUB1', 'BUB1B'})})

Hand-curated name pairs that group_name_homologs() must never collapse together, overriding the prefix heuristic. BUB1 / BUB1B are distinct genes (BUB1B is BUBR1) whose B suffix is not the receptor R case caught generically.

Type:

frozenset[frozenset[str]]

mkt.schema.utils.TQDM_BAR_FORMAT = '{l_bar}{bar}| {n:,}/{total:,} [{elapsed}<{remaining}, {rate_fmt}{postfix}]'

Default tqdm bar format with comma-separated thousands in counts.

mkt.schema.utils.adjudicate_kinase_group(str_kinase: str, bool_lipid: bool = True) → str | None

Adjudicates the kinase group for a given kinase.

Parameters:

• str_kinase (str) – The name of the kinase (e.g., “PIK3CA”).

• bool_lipid (bool, optional) – Flag to indicate if lipid kinases should be classified as “Lipid” group, by default True.

Returns:

The adjudicated kinase group (e.g., “Lipid”, “TK”, “CMGC”), or None if the kinase is not found.

Return type:

str | None

mkt.schema.utils.group_name_homologs(names: list[str], min_prefix: int = 3, show_count: bool = True) → list[tuple[str, list[str]]]

Collapse prefix-homologous kinase names into compact labeled groups.

Names are grouped when they share a common base stem of at least min_prefix characters, each member’s remaining variant is a single letter or a pure number (so distinct subfamilies such as EPHA10 / EPHB6 stay apart), and they carry the same multi-domain suffix (see split_domain_suffix()). The stem is trimmed back off any mid-number boundary so a shorter member number is not split out of a longer one (NEK1 vs NEK10/NEK11), and numeric variants are sorted numerically. Each group is labeled by factoring out the stem, e.g. ["JAK1_1", "JAK2_1", "JAK3_1"] -> ("JAK1/2/3_1 (3)", [...]) or ["NEK1", "NEK10", "NEK2"] -> ("NEK1/2/10", [...]). The default min_prefix of 3 keeps coincidental two-character matches apart (e.g. the unrelated ATM, ATR). A complete gene name and its stem + "R" receptor paralog are never merged (e.g. INSR / INSRR stay apart), and any pair in SET_HOMOLOG_DO_NOT_MERGE is held apart by hand (BUB1 / BUB1B).

Parameters:

• names (list[str]) – Kinase names to group. Order is not significant: names are sorted by (domain_suffix, name) internally so homologs are adjacent.

• min_prefix (int, optional) – Minimum shared base-stem length required to merge, by default 3.

• show_count (bool, optional) – Append a " (N)" member count to each merged group’s label, by default True.

Returns:

(label, members) pairs in sorted order; a singleton is (name, [name]).

Return type:

list[tuple[str, list[str]]]

mkt.schema.utils.random_uuid()

Generate a random UUID that allows to set a seed.

Returns:

A random UUID as a string.

Return type:

str

mkt.schema.utils.return_kinase_gene_set(dict_kinase: dict | None = None) → set[str]

Return the set of kinase HGNC gene symbols for gene-level membership tests.

Multi-kinase-domain genes are keyed with a _1 / _2 domain suffix (see split_domain_suffix()); those are collapsed to the base gene symbol (e.g. JAK1_1 / JAK1_2 -> JAK1) so a symbol like "JAK1" is not missed when testing membership against cohort gene symbols.

Parameters:

dict_kinase (dict | None) – Mapping keyed by kinase name (optionally carrying a _<digit> domain suffix). If None, the canonical DICT_KINASE is deserialized.

Returns:

Base HGNC gene symbols of every kinase.

Return type:

set[str]

mkt.schema.utils.rgetattr(obj, attr, *args)

Get attribute from object recursively.

Parameters:

• obj (Any) – Object to get attribute from.

• attr (str) – Attribute to get.

• *args (Any) – Any additional arguments to pass to getattr.

Returns:

Value of attribute if found.

Return type:

Any

mkt.schema.utils.rsetattr(obj, attr, val)

Set attribute from object recursively.

Parameters:

• obj (Any) – Object to get attribute from.

• attr (str) – Attribute to get.

• val (Any) – Value to set attribute to.

Returns:

Value of attribute if found, otherwise default value.

Return type:

Any

mkt.schema.utils.split_domain_suffix(name: str) → tuple[str, str]

Split a trailing multi-domain index suffix off a kinase name.

Multi-domain kinases are represented with a _1 / _2 suffix denoting the individual kinase domains (e.g. "JAK1_1" is the first kinase domain of JAK1).

Parameters:

name (str) – Kinase name, optionally carrying a _<digit> domain suffix.

Returns:

(base, suffix) where suffix is "_<digit>" or "" (e.g. "JAK1_1" -> ("JAK1", "_1"), "BTK" -> ("BTK", "")).

Return type:

tuple[str, str]