reverse_complement_generation.py

from typing import Dict
class InvalidNucleotideError(Exception):
    """Raised when an invalid nucleotide is encountered."""
    pass

#Complement Mapping (IUPAC codes)
complement_map= {
    #standard bases
    'A': 'T', 'T': 'A', 'G': 'C', 'C': 'G',
    'a': 't', 't': 'a', 'g': 'c', 'c': 'g',

    # Degenerate Nucleotides
    'R': 'Y', 'Y': 'R', 'S': 'S', 'W': 'W', 'K': 'M', 'M': 'K',
    'H': 'D', 'D': 'H', 'V': 'B', 'B': 'V', 'N': 'N', 'X': 'X',
    'r': 'y', 'y': 'r', 's': 's', 'w': 'w', 'k': 'm', 'm': 'k',
    'h': 'd', 'd': 'h', 'v': 'b', 'b': 'v', 'n': 'n', 'x': 'x'
}

valid_iupac_nucleotides= set(complement_map.keys())

def generate_reverse_complement(sequence: str,
                                orientation: str = "5'-3'",
                                case_sensitive: bool = False
                                )->str:
    """
        Builds robust reverse complement functionality.

        Args:
            sequence: The input DNA sequence string.
            orientation: The desired output orientation.
                         Options: "5'-3'" (default, the standard result) or
                                  "3'-5'" (the complement before reversal).
            case_sensitive: If True, maintains the original case of the output bases
                            (useful for masked sequences). Default is False (uppercase).

        Returns:
            The reverse complement sequence string in the specified orientation.

        Raises:
            InvalidNucleotideError: If the sequence contains non-IUPAC characters.
            ValueError: If an invalid orientation is provided.
        """
    if not case_sensitive:
        working_sequence = sequence.upper()
        complement_map_used={k.upper():v.upper() for k, v in complement_map.items()}
    else:
        working_sequence = sequence
        complement_map_used= complement_map

    invalid_characters= set(working_sequence) - set(complement_map_used.keys())
    if invalid_characters:
        raise InvalidNucleotideError(
            f"Invalid nucleotides present: {",".join(invalid_characters)}"
        )

    translation_table= str.maketrans(complement_map_used)
    complement= working_sequence.translate(translation_table)

    orientation= orientation.strip().lower()
    if orientation == "5'-3'":
        return complement[::-1]
    elif orientation == "3'-5'":
        return complement
    else:
        raise ValueError(f"Invalid orientation:{orientation}")