コード例 #1
0
ファイル: 20130221a.py プロジェクト: argriffing/xgcode
def get_response_content(fs):

    # define the amount of time we will search
    nseconds = 5

    # set up print options
    np.set_printoptions(
            linewidth=1000000,
            threshold=1000000,
            )
    out = StringIO()

    # set up some process parameters
    blink_birth = fs.blink_birth
    blink_death = fs.blink_death
    rate_01 = fs.rate_01
    rate_12 = fs.rate_12
    rate_20 = fs.rate_20

    # define the original rate matrix
    pre_Q = np.array([
        [0, rate_01, 0],
        [0, 0, rate_12],
        [rate_20, 0, 0],
        ], dtype=float)

    # construct the derived rate matrix
    pre_blink = binarytolerance.blinkize(pre_Q, blink_birth, blink_death)

    # check stationary distributions
    Q = binarytolerance.pre_Q_to_Q(pre_Q)
    W, V = scipy.linalg.eig(Q.T)
    v = V[:, np.argmin(np.abs(W))]
    v /= np.sum(v)
    #print >> out, Q
    #print >> out, W
    print >> out, 'original process stationary distribution:'
    print >> out, v
    print >> out

    Q_blink = binarytolerance.pre_Q_to_Q(pre_blink)
    W, V = scipy.linalg.eig(Q_blink.T)
    v = V[:, np.argmin(np.abs(W))]
    v /= np.sum(v)
    #print >> out, Q_blink
    #print >> out, W
    #print >> out, v
    #print >> out

    n = pre_Q.shape[0]
    expo = binarytolerance.get_expanded_states(n)
    x = np.zeros(n)
    for i, (perm, mask) in enumerate(expo):
        x[perm[0]] += v[i]
    print >> out, 'blink-ized stationary distribution:'
    print >> out, x
    print >> out

    # show the result
    return out.getvalue()
コード例 #2
0
def get_response_content(fs):

    # define the amount of time we will search
    nseconds = 5

    # set up print options
    np.set_printoptions(
        linewidth=1000000,
        threshold=1000000,
    )
    out = StringIO()

    # set up some process parameters
    blink_birth = fs.blink_birth
    blink_death = fs.blink_death
    rate_01 = fs.rate_01
    rate_12 = fs.rate_12
    rate_20 = fs.rate_20

    # define the original rate matrix
    pre_Q = np.array([
        [0, rate_01, 0],
        [0, 0, rate_12],
        [rate_20, 0, 0],
    ],
                     dtype=float)

    # construct the derived rate matrix
    pre_blink = binarytolerance.blinkize(pre_Q, blink_birth, blink_death)

    # check stationary distributions
    Q = binarytolerance.pre_Q_to_Q(pre_Q)
    W, V = scipy.linalg.eig(Q.T)
    v = V[:, np.argmin(np.abs(W))]
    v /= np.sum(v)
    #print >> out, Q
    #print >> out, W
    print >> out, 'original process stationary distribution:'
    print >> out, v
    print >> out

    Q_blink = binarytolerance.pre_Q_to_Q(pre_blink)
    W, V = scipy.linalg.eig(Q_blink.T)
    v = V[:, np.argmin(np.abs(W))]
    v /= np.sum(v)
    #print >> out, Q_blink
    #print >> out, W
    #print >> out, v
    #print >> out

    n = pre_Q.shape[0]
    expo = binarytolerance.get_expanded_states(n)
    x = np.zeros(n)
    for i, (perm, mask) in enumerate(expo):
        x[perm[0]] += v[i]
    print >> out, 'blink-ized stationary distribution:'
    print >> out, x
    print >> out

    # show the result
    return out.getvalue()
コード例 #3
0
ファイル: 20130219a.py プロジェクト: argriffing/xgcode
def get_response_content(fs):

    # define the amount of time we will search
    nseconds = 5

    # set up print options
    np.set_printoptions(linewidth=1000000, threshold=1000000)
    out = StringIO()

    # set up some process parameters
    blink_birth = fs.blink_birth
    blink_death = fs.blink_death
    if blink_birth < 0:
        raise Exception
    if blink_death < 0:
        raise Exception

    # define the original rate matrix
    # pre_Q = np.exp(np.random.randn(3, 3))
    pre_Q = binarytolerance.sample_reversible_pre_Q(4)
    """
    pre_Q = np.array([
        [0.0, 0.5],
        [0.1, 0.0],
        ], dtype=float)
    pre_Q = np.array([
        [0.0, 0.3, 0.1],
        [0.1, 0.0, 0.3],
        [0.1, 0.1, 0.0],
        ], dtype=float)
    pre_Q = np.array([
        [0.0, 3.0, 0.1],
        [0.1, 0.0, 3.0],
        [0.1, 0.1, 0.0],
        ], dtype=float)
    pre_Q = np.array([
        [0.00, 6.00, 0.00],
        [0.00, 0.00, 6.00],
        [0.01, 0.00, 0.00],
        ], dtype=float)
    """

    # construct the derived rate matrix
    pre_blink = binarytolerance.blinkize(pre_Q, blink_birth, blink_death)

    # check stationary distributions
    Q = binarytolerance.pre_Q_to_Q(pre_Q)
    W, V = scipy.linalg.eig(Q.T)
    v = V[:, np.argmin(np.abs(W))]
    v /= np.sum(v)
    v_mutational = v
    print >> out, Q
    print >> out, W
    print >> out, "mutational process amino acid stationary distribution:"
    print >> out, v
    print >> out, np.sum(v)
    print >> out

    Q_blink = binarytolerance.pre_Q_to_Q(pre_blink)
    W, V = scipy.linalg.eig(Q_blink.T)
    v = V[:, np.argmin(np.abs(W))]
    v /= np.sum(v)
    print >> out, Q_blink
    print >> out, W
    print >> out, "full stationary distribution:"
    print >> out, v
    print >> out, np.sum(v)
    print >> out

    n = pre_Q.shape[0]
    expo = binarytolerance.get_expanded_states(n)
    x = np.zeros(n)
    for i, (perm, mask) in enumerate(expo):
        x[perm[0]] += v[i]
    print >> out, "marginal amino acid stationary distribution:"
    print >> out, x
    print >> out, np.sum(x)
    print >> out

    # Attempt to construct the full stationary distribution as the product
    # of the mutational process amino acid stationary probability
    # and a binomial stationary probability of the blinking process.
    n = pre_Q.shape[0]
    p_on = blink_birth / (blink_birth + blink_death)
    p_off = blink_death / (blink_birth + blink_death)
    blink_factorial = math.factorial(n - 1)
    expo = binarytolerance.get_expanded_states(n)
    x = np.ones(len(expo), dtype=float)
    for i, (perm, mask) in enumerate(expo):
        # Account for the amino acid residue mutational probability.
        x[i] *= v_mutational[perm[0]]
        # Account for the blinking permutations.
        x[i] /= blink_factorial
        # Account for the binomial probability of blinks.
        n_on = sum(mask[1:])
        n_off = (n - 1) - n_on
        # x[i] *= scipy.special.binom(n_on, p_on)
        # x[i] *= scipy.stats.binom.pmf(n_on, n-1, p_on)
        x[i] *= p_on ** n_on
        x[i] *= p_off ** n_off
    print >> out, "attempt to construct the full stationary distribution:"
    print >> out, x
    print >> out, np.sum(x)
    print >> out

    # show the result
    return out.getvalue()
コード例 #4
0
ファイル: 20130219a.py プロジェクト: BIGtigr/xgcode
def get_response_content(fs):

    # define the amount of time we will search
    nseconds = 5

    # set up print options
    np.set_printoptions(
        linewidth=1000000,
        threshold=1000000,
    )
    out = StringIO()

    # set up some process parameters
    blink_birth = fs.blink_birth
    blink_death = fs.blink_death
    if blink_birth < 0:
        raise Exception
    if blink_death < 0:
        raise Exception

    # define the original rate matrix
    #pre_Q = np.exp(np.random.randn(3, 3))
    pre_Q = binarytolerance.sample_reversible_pre_Q(4)
    """
    pre_Q = np.array([
        [0.0, 0.5],
        [0.1, 0.0],
        ], dtype=float)
    pre_Q = np.array([
        [0.0, 0.3, 0.1],
        [0.1, 0.0, 0.3],
        [0.1, 0.1, 0.0],
        ], dtype=float)
    pre_Q = np.array([
        [0.0, 3.0, 0.1],
        [0.1, 0.0, 3.0],
        [0.1, 0.1, 0.0],
        ], dtype=float)
    pre_Q = np.array([
        [0.00, 6.00, 0.00],
        [0.00, 0.00, 6.00],
        [0.01, 0.00, 0.00],
        ], dtype=float)
    """

    # construct the derived rate matrix
    pre_blink = binarytolerance.blinkize(pre_Q, blink_birth, blink_death)

    # check stationary distributions
    Q = binarytolerance.pre_Q_to_Q(pre_Q)
    W, V = scipy.linalg.eig(Q.T)
    v = V[:, np.argmin(np.abs(W))]
    v /= np.sum(v)
    v_mutational = v
    print >> out, Q
    print >> out, W
    print >> out, 'mutational process amino acid stationary distribution:'
    print >> out, v
    print >> out, np.sum(v)
    print >> out

    Q_blink = binarytolerance.pre_Q_to_Q(pre_blink)
    W, V = scipy.linalg.eig(Q_blink.T)
    v = V[:, np.argmin(np.abs(W))]
    v /= np.sum(v)
    print >> out, Q_blink
    print >> out, W
    print >> out, 'full stationary distribution:'
    print >> out, v
    print >> out, np.sum(v)
    print >> out

    n = pre_Q.shape[0]
    expo = binarytolerance.get_expanded_states(n)
    x = np.zeros(n)
    for i, (perm, mask) in enumerate(expo):
        x[perm[0]] += v[i]
    print >> out, 'marginal amino acid stationary distribution:'
    print >> out, x
    print >> out, np.sum(x)
    print >> out

    # Attempt to construct the full stationary distribution as the product
    # of the mutational process amino acid stationary probability
    # and a binomial stationary probability of the blinking process.
    n = pre_Q.shape[0]
    p_on = blink_birth / (blink_birth + blink_death)
    p_off = blink_death / (blink_birth + blink_death)
    blink_factorial = math.factorial(n - 1)
    expo = binarytolerance.get_expanded_states(n)
    x = np.ones(len(expo), dtype=float)
    for i, (perm, mask) in enumerate(expo):
        # Account for the amino acid residue mutational probability.
        x[i] *= v_mutational[perm[0]]
        # Account for the blinking permutations.
        x[i] /= blink_factorial
        # Account for the binomial probability of blinks.
        n_on = sum(mask[1:])
        n_off = (n - 1) - n_on
        #x[i] *= scipy.special.binom(n_on, p_on)
        #x[i] *= scipy.stats.binom.pmf(n_on, n-1, p_on)
        x[i] *= p_on**n_on
        x[i] *= p_off**n_off
    print >> out, 'attempt to construct the full stationary distribution:'
    print >> out, x
    print >> out, np.sum(x)
    print >> out

    # show the result
    return out.getvalue()