DNA

# DNA.py
# Heidi Heffelfinger
# October 17, 2012
# Learn to Program - Coursera - University of Toronto
# Week 4 Assignment
# Python 3.2

# This program examines DNA sequences, checks for validity, and makes
# modifications to DNA sequences.

def get_length(dna):
    ''' (str) -> int

    Return the length of the DNA sequence dna.

    >>> get_length('ATCGAT')
    6
    >>> get_length('ATCG')
    4
    >>> get_length('CGATAGCT')
    8
    >>> get_length('CG')
    2
    '''
    return len(dna)


def is_longer(dna1, dna2):
    ''' (str, str) -> bool

    Return True if and only if DNA sequence dna1 is longer than DNA sequence
    dna2.

    >>> is_longer('ATCG', 'AT')
    True
    >>> is_longer('ATCG', 'ATCGGA')
    False
    >>> is_longer('ATCG', 'CGATAGCT')
    False
    >>> is_longer('ATCGAT', 'CGA')
    True
    '''
    return get_length(dna1) > get_length(dna2)


def count_nucleotides(dna, nucleotide):
    ''' (str, str) -> int

    Return the number of occurrences of nucleotide in the DNA sequence dna.

    >>> count_nucleotides('ATCGGC', 'G')
    2
    >>> count_nucleotides('ATCTA', 'G')
    0
    >>> count_nucleotides('CGATAGCT', 'T')
    2
    >>> count_nucleotides('ATCGGA', 'C')
    1
    '''
    return dna.count(nucleotide)


def contains_sequence(dna1, dna2):
    ''' (str, str) -> bool

    Return True if and only if DNA sequence dna2 occurs in the DNA sequence
    dna1.

    >>> contains_sequence('ATCGGC', 'GG')
    True
    >>> contains_sequence('ATCGGC', 'GT')
    False
    >>> contains_sequence('ATCGGC', 'CG')
    True
    >>> contains_sequence('ATCGGC', 'AT')
    True
    >>> contains_sequence('ATCGGC', 'AG')
    False
    '''
    return dna2 in dna1

def is_valid_sequence(dna):
    ''' (str) -> bool

    Return True if dna sequence is valid, containing only nucleotide
    characters: 'A', 'T', 'C' and 'G'.

    >>> is_valid_sequence('ATCGGC')
    True
    >>> is_valid_sequence('ATcGGC')
    False
    >>> is_valid_sequence('CGATAGCT')
    True
    >>> is_valid_sequence('ABCDEFG')
    False
    >>> is_valid_sequence('atcggc')
    False
    >>> is_valid_sequence('CGATAGCT')
    True
    >>> is_valid_sequence('A')
    True
    '''
    dna_nucleotides = 'ATCG'
    for nucleotide in dna:
        if nucleotide not in dna_nucleotides:
            return False
    return True

def insert_sequence(dna1, dna2, insertion):
    '''(str, str, int) -> str

    Returns DNA sequence obtained by inserting dna2 into dna1 at insertion point.

    >>> insert_sequence('CTGATAGCT', 'ATCGGC', 2)
    CTATCGGCGATAGCT
    >>> insert_sequence('ATCGGC', 'CG', 4)
    ATCGCGGC
    >>> insert_sequence('ATCGGC', 'TA', 1)
    ATATCGGC
    '''

    new_dna = dna1[:insertion] + dna2 + dna1[insertion:]
    return new_dna

def get_complement(nucleotide):
    '''(str) -> str

    Returns complement (A, T, C, G) of nucleotide (A, T, C, G) where A is
    complement of T (& vice-versa) and C is complement of G (& vice versa).

    Precondition - nucleotide must be A, T, C, or G.

    >>> get_complement('A')
    T
    >>> get_complement('T')
    A
    >>> get_complement('C')
    G
    >>> get_complement('G')
    C
    '''
    if nucleotide == 'A':
        return 'T'
    elif nucleotide == 'T':
        return 'A'
    elif nucleotide == 'C':
        return 'G'
    elif nucleotide == 'G':
        return 'C'

def get_complementary_sequence(dna):
    '''(str) -> str

    Returns DNA sequence that is complementary to given DNA sequence of dna.

    >>> get_complementary_sequence('ACGTACG')
    TGCATGC
    >>> get_complementary_sequence('CGATAGCT')
    GCTATCGA
    >>> get_complementary_sequence('ATCG')
    TAGC
    >>> get_complementary_sequence('ATCTA')
    TAGAT
    '''
    complement = ''
    for nucleotide in dna:
        complement += get_complement(nucleotide)
    return complement