utils/fasta_parser.py

from typing import List, Tuple

VALID_AA = set("ACDEFGHIKLMNPQRSTVWYXBUZO")


def parse_fasta(fasta_text: str) -> List[Tuple[str, str]]:
    """
    Parse a FASTA string and return list of (protein_id, sequence).
    Raises ValueError on malformed or empty input.
    """
    proteins: List[Tuple[str, str]] = []
    current_id: str = ""
    current_seq: List[str] = []

    for line in fasta_text.strip().splitlines():
        line = line.strip()
        if not line:
            continue
        if line.startswith(">"):
            if current_id:
                seq = "".join(current_seq).upper()
                _validate_sequence(current_id, seq)
                proteins.append((current_id, seq))
            current_id = line[1:].split()[0]   # first token after ">"
            current_seq = []
        else:
            current_seq.append(line)

    # last protein
    if current_id:
        seq = "".join(current_seq).upper()
        _validate_sequence(current_id, seq)
        proteins.append((current_id, seq))

    if not proteins:
        raise ValueError("No valid protein sequences found in the FASTA file.")

    return proteins


def _validate_sequence(protein_id: str, seq: str) -> None:
    if not seq:
        raise ValueError(f"Protein '{protein_id}' has an empty sequence.")
    invalid = set(seq) - VALID_AA
    if invalid:
        raise ValueError(
            f"Protein '{protein_id}' contains invalid amino acid characters: {invalid}"
        )


def sliding_window(sequence: str, window_size: int, stride: int) -> List[str]:
    """
    Split a long sequence into overlapping windows.
    If the sequence fits in one window, returns it as-is.
    """
    if len(sequence) <= window_size:
        return [sequence]

    windows = []
    start = 0
    while start < len(sequence):
        end = min(start + window_size, len(sequence))
        windows.append(sequence[start:end])
        if end == len(sequence):
            break
        start += stride
    return windows