dna_string_manipulation.py

from typing import List, Optional

valid_nucleotides = {"A","T","G","C","a","t","c","g"}

def to_uppercase(sequence:str)->str:
    """
    Converts all nucleotides to upper case
    eg- "atcg" -> "ATCG
    """
    return sequence.upper()

def to_lowercase(sequence:str)->str:
    """"
    Converts all nucleotides to lower case
    eg- "atcg" -> "ATCG
    """
    return sequence.lower()

def remove_non_nucleotides(sequence:str,valid_characters:set= valid_nucleotides)->str:
    """"
    Removes non-nucleotide characters from sequence
    eg- "ATTCG!XBYTTGC" -> "ATTCGTTGC"
    """
    cleaned_sequence="".join(char for char in sequence if char in valid_characters)
    return cleaned_sequence

def split_into_codons(sequence:str, reading_frame:int=1)->List[str]:
    """"
    Splits sequence into codons (groups of 3 nucleotides)
    Reading frame determines where to start (1= index 0, 2= index 1, 3= index 2)
    Any leftover nucleotides (<3 bases) will be ignored
    """
    sequence = to_uppercase(sequence)
    start_index= reading_frame-1

    if not 0<= start_index< 3:
        raise ValueError('start_index must be between 1 and 3')

    codons=[]
    for i in range (start_index , len(sequence), 3):
        codon= sequence[i:i+3]
        if(len(codon)==3):
            codons.append(codon)
    return codons

def merge_dna_fragments(fragments:List[str], separator:Optional[str]=" ") -> str:
    """"
    Merges a list of DNA fragments into a single sequence
    Optionally insert a separator between fragments
    """
    cleaned_fragments= [str(f) for f in fragments]
    return separator.join(cleaned_fragments)