Prozor¶

Prozor is a typed Python library for matching peptide sequences to proteins and inferring deterministic parsimonious protein groups.

It provides two connected building blocks:

• backend-neutral Aho--Corasick matching over in-memory or streaming protein records; and
• unique peptide--protein edges with deterministic greedy-parsimony protein inference.

Get started API reference

Why Prozor?¶

• Complete matching: nested, repeated, and overlapping peptide occurrences are retained with half-open coordinates.
• Streaming input: protein records can be consumed from any one-pass iterable; Prozor does not prescribe a FASTA parser.
• Auditable backends: results record both the requested backend and the concrete implementation selected at runtime; Rust is the default and pure Python remains the fallback.
• Deterministic inference: equivalent protein groups and tie-breaking use a stable ordering.
• Small boundary: the core does not depend on pandas, AnnData, MuData, MuLink, workflow engines, or consumer CLIs.

Minimal workflow¶

from dataclasses import dataclass

from prozor.api import annotate_peptides_streaming, greedy_parsimony


@dataclass(frozen=True, slots=True)
class Protein:
    id: str
    sequence: str


matches = annotate_peptides_streaming(
    ["PEPTIDE", "SEQUENCE"],
    [
        Protein("P1", "MYPEPTIDESEQUENCE"),
        Protein("P2", "XXSEQUENCEXX"),
    ],
)

edges = {(match.peptide, match.protein_id) for match in matches}
protein_groups = greedy_parsimony(edges)
print(protein_groups.to_dict())

Project status¶

The Python package is derived from the algorithmic behavior of the R prozor reference package. The matching and inference core is covered by pure-Python/Rust backend-equivalence tests, strict Pyright, Ruff, branch coverage, dependency validation, and distribution builds.