base.py

#!/usr/bin/env python
from __future__ import division

# ----------------------------------------------------------------------------
# Copyright (c) 2013--, scikit-bio development team.
#
# Distributed under the terms of the Modified BSD License.
#
# The full license is in the file COPYING.txt, distributed with this software.
# ----------------------------------------------------------------------------

import numpy as np
from scipy.special import gammaln
from scipy.optimize import fmin_powell, minimize_scalar

from skbio.math.subsample import subsample


def _validate(counts, suppress_cast=False):
    """Validate and convert input to an acceptable counts vector type.

    Note: may not always return a copy of `counts`!

    """
    counts = np.asarray(counts)

    if not suppress_cast:
        counts = counts.astype(int, casting='safe', copy=False)

    if counts.ndim != 1:
        raise ValueError("Only 1-D vectors are supported.")
    elif (counts < 0).any():
        raise ValueError("Counts vector cannot contain negative values.")

    return counts


def berger_parker_d(counts):
    """Calculate Berger-Parker dominance.

    Berger-Parker dominance is defined as the fraction of the sample that
    belongs to the most abundant OTUs.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        Berger-Parker dominance.

    Notes
    -----
    Berger-Parker dominance is defined in [1]_. The implementation here is
    based on the description given in the SDR-IV online manual [2]_.

    References
    ----------
    .. [1] Berger & Parker (1970). SDR-IV online help.
    .. [2] http://www.pisces-conservation.com/sdrhelp/index.html

    """
    counts = _validate(counts)
    return counts.max() / counts.sum()


def brillouin_d(counts):
    """Calculate Brillouin index of alpha diversity.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        Brillouin index.

    Notes
    -----
    The implementation here is based on the description given in the SDR-IV
    online manual [1]_.

    References
    ----------
    .. [1] http://www.pisces-conservation.com/sdrhelp/index.html

    """
    counts = _validate(counts)
    nz = counts[counts.nonzero()]
    n = nz.sum()
    return (gammaln(n + 1) - gammaln(nz + 1).sum()) / n


def dominance(counts):
    """Calculate dominance.

    Dominance is defined as

    .. math::

       \\sum{p_i^2}

    where :math:`p_i` is the proportion of the entire community that OTU
    :math:`i` represents.

    Dominance can also be defined as 1 - Simpson's index. It ranges between
    0 and 1.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        Dominance.

    See Also
    --------
    simpson

    Notes
    -----
    The implementation here is based on the description given in [1]_.

    References
    ----------
    .. [1] http://folk.uio.no/ohammer/past/diversity.html

    """
    counts = _validate(counts)
    freqs = counts / counts.sum()
    return (freqs * freqs).sum()


def doubles(counts):
    """Calculate number of double occurrences (doubletons).

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    int
        Doubleton count.

    """
    counts = _validate(counts)
    return (counts == 2).sum()


def enspie(counts):
    """Calculate ENS_pie alpha diversity measure.

    ENS_pie is equivalent to ``1 / dominance``.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        ENS_pie alpha diversity measure.

    See Also
    --------
    dominance

    Notes
    -----
    ENS_pie is defined in [1]_.

    References
    ----------
    .. [1] Chase and Knight (2013). "Scale-dependent effect sizes of ecological
       drivers on biodiversity: why standardised sampling is not enough".
       Ecology Letters, Volume 16, Issue Supplement s1, pgs 17-26.

    """
    counts = _validate(counts)
    return 1 / dominance(counts)


def equitability(counts, base=2):
    """Calculate equitability (Shannon index corrected for number of OTUs).

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.
    base : scalar, optional
        Logarithm base to use in the calculations.

    Returns
    -------
    double
        Measure of equitability.

    See Also
    --------
    shannon

    Notes
    -----
    The implementation here is based on the description given in the SDR-IV
    online manual [1]_.

    References
    ----------
    .. [1] http://www.pisces-conservation.com/sdrhelp/index.html

    """
    counts = _validate(counts)
    numerator = shannon(counts, base)
    denominator = np.log(observed_otus(counts)) / np.log(base)
    return numerator / denominator


def esty_ci(counts):
    """Calculate Esty's CI.

    Esty's CI is defined as

    .. math::

       F_1/N \\pm z\\sqrt{W}

    where :math:`F_1` is the number of singleton OTUs, :math:`N` is the total
    number of individuals (sum of abundances for all OTUs), and :math:`z` is a
    constant that depends on the targeted confidence and based on the normal
    distribution.

    :math:`W` is defined as

    .. math::

       \\frac{F_1(N-F_1)+2NF_2}{N^3}

    where :math:`F_2` is the number of doubleton OTUs.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    tuple
        Esty's confidence interval as ``(lower_bound, upper_bound)``.

    Notes
    -----
    Esty's CI is defined in [1]_. :math:`z` is hardcoded for a 95% confidence
    interval.

    References
    ----------
    .. [1] Esty, W. W. (1983). "A normal limit law for a nonparametric
       estimator of the coverage of a random sample". Ann Statist 11: 905-912.

    """
    counts = _validate(counts)

    f1 = singles(counts)
    f2 = doubles(counts)
    n = counts.sum()
    z = 1.959963985
    W = (f1 * (n - f1) + 2 * n * f2) / (n ** 3)

    return f1 / n - z * np.sqrt(W), f1 / n + z * np.sqrt(W)


def fisher_alpha(counts):
    """Calculate Fisher's alpha.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        Fisher's alpha.

    Raises
    ------
    RuntimeError
        If the optimizer fails to converge (error > 1.0).

    Notes
    -----
    The implementation here is based on the description given in the SDR-IV
    online manual [1]_. Uses ``scipy.optimize.minimize_scalar`` to find
    Fisher's alpha.

    References
    ----------
    .. [1] http://www.pisces-conservation.com/sdrhelp/index.html

    """
    counts = _validate(counts)
    n = counts.sum()
    s = observed_otus(counts)

    def f(alpha):
        return (alpha * np.log(1 + (n / alpha)) - s) ** 2

    # Temporarily silence RuntimeWarnings (invalid and division by zero) during
    # optimization in case invalid input is provided to the objective function
    # (e.g. alpha=0).
    orig_settings = np.seterr(divide='ignore', invalid='ignore')
    try:
        alpha = minimize_scalar(f).x
    finally:
        np.seterr(**orig_settings)

    if f(alpha) > 1.0:
        raise RuntimeError("Optimizer failed to converge (error > 1.0), so "
                           "could not compute Fisher's alpha.")
    return alpha


def goods_coverage(counts):
    """Calculate Good's coverage of counts.

    Good's coverage estimator is defined as

    .. math::

       1-\\frac{F_1}{N}

    where :math:`F_1` is the number of singleton OTUs and :math:`N` is the
    total number of individuals (sum of abundances for all OTUs).

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        Good's coverage estimator.

    """
    counts = _validate(counts)
    f1 = singles(counts)
    N = counts.sum()
    return 1 - (f1 / N)


def heip_e(counts):
    """Calculate Heip's evenness measure.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        Heip's evenness measure.

    Notes
    -----
    The implementation here is based on the description in [1]_.

    References
    ----------
    .. [1] Heip, C. 1974. A new index measuring evenness. J. Mar. Biol. Ass.
       UK., 54, 555-557.

    """
    counts = _validate(counts)
    return ((np.exp(shannon(counts, base=np.e)) - 1) /
            (observed_otus(counts) - 1))


def kempton_taylor_q(counts, lower_quantile=0.25, upper_quantile=0.75):
    """Calculate Kempton-Taylor Q index of alpha diversity.

    Estimates the slope of the cumulative abundance curve in the interquantile
    range. By default, uses lower and upper quartiles, rounding inwards.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.
    lower_quantile : float, optional
        Lower bound of the interquantile range. Defaults to lower quartile.
    upper_quantile : float, optional
        Upper bound of the interquantile range. Defaults to upper quartile.

    Returns
    -------
    double
        Kempton-Taylor Q index of alpha diversity.

    Notes
    -----
    The index is defined in [1]_. The implementation here is based on the
    description given in the SDR-IV online manual [2]_.

    The implementation provided here differs slightly from the results given in
    Magurran 1998. Specifically, we have 14 in the numerator rather than 15.
    Magurran recommends counting half of the OTUs with the same # counts as the
    point where the UQ falls and the point where the LQ falls, but the
    justification for this is unclear (e.g. if there were a very large # OTUs
    that just overlapped one of the quantiles, the results would be
    considerably off). Leaving the calculation as-is for now, but consider
    changing.

    References
    ----------
    .. [1] Kempton, R. A. and Taylor, L. R. (1976) Models and statistics for
       species diversity. Nature, 262, 818-820.
    .. [2] http://www.pisces-conservation.com/sdrhelp/index.html

    """
    counts = _validate(counts)
    n = len(counts)
    lower = int(np.ceil(n * lower_quantile))
    upper = int(n * upper_quantile)
    sorted_counts = np.sort(counts)
    return (upper - lower) / np.log(sorted_counts[upper] /
                                    sorted_counts[lower])


def margalef(counts):
    """Calculate Margalef's richness index.

    Assumes log accumulation.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        Margalef's richness index.

    Notes
    -----
    Based on the description in [1]_.

    References
    ----------
    .. [1] Magurran, A E 2004. Measuring biological diversity. Blackwell. pp.
       76-77.

    """
    counts = _validate(counts)
    return (observed_otus(counts) - 1) / np.log(counts.sum())


def mcintosh_d(counts):
    """Calculate McIntosh dominance index D.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        McIntosh dominance index D.

    See Also
    --------
    mcintosh_e

    Notes
    -----
    The index was proposed in [1]_. The implementation here is based on the
    description given in the SDR-IV online manual [2]_.

    References
    ----------
    .. [1] McIntosh, R. P. 1967 An index of diversity and the relation of
       certain concepts to diversity. Ecology 48, 1115-1126.
    .. [2] http://www.pisces-conservation.com/sdrhelp/index.html

    """
    counts = _validate(counts)
    u = np.sqrt((counts * counts).sum())
    n = counts.sum()
    return (n - u) / (n - np.sqrt(n))


def mcintosh_e(counts):
    """Calculate McIntosh's evenness measure E.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        McIntosh evenness measure E.

    See Also
    --------
    mcintosh_d

    Notes
    -----
    The implementation here is based on the description given in [1]_, *NOT*
    the one in the SDR-IV online manual, which is wrong.

    References
    ----------
    .. [1] Heip & Engels 1974 p 560.

    """
    counts = _validate(counts)
    numerator = np.sqrt((counts * counts).sum())
    n = counts.sum()
    s = observed_otus(counts)
    denominator = np.sqrt((n - s + 1) ** 2 + s - 1)
    return numerator / denominator


def menhinick(counts):
    """Calculate Menhinick's richness index.

    Assumes square-root accumulation.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        Menhinick's richness index.

    Notes
    -----
    Based on the description in [1]_.

    References
    ----------
    .. [1] Magurran, A E 2004. Measuring biological diversity. Blackwell. pp.
       76-77.

    """
    counts = _validate(counts)
    return observed_otus(counts) / np.sqrt(counts.sum())


def michaelis_menten_fit(counts, num_repeats=1, params_guess=None):
    """Calculate Michaelis-Menten fit to rarefaction curve of observed OTUs.

    The Michaelis-Menten equation is defined as

    .. math::

       S=\\frac{nS_{max}}{n+B}

    where :math:`n` is the number of individuals and :math:`S` is the number of
    OTUs. This function estimates the :math:`S_{max}` parameter.

    The fit is made to datapoints for :math:`n=1,2,...,N`, where :math:`N` is
    the total number of individuals (sum of abundances for all OTUs).
    :math:`S` is the number of OTUs represented in a random sample of :math:`n`
    individuals.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.
    num_repeats : int, optional
        The number of times to perform rarefaction (subsampling without
        replacement) at each value of :math:`n`.
    params_guess : tuple, optional
        Initial guess of :math:`S_{max}` and :math:`B`. If ``None``, default
        guess for :math:`S_{max}` is :math:`S` (as :math:`S_{max}` should
        be >= :math:`S`) and default guess for :math:`B` is ``round(N / 2)``.

    Returns
    -------
    S_max : double
        Estimate of the :math:`S_{max}` parameter in the Michaelis-Menten
        equation.

    See Also
    --------
    skbio.math.subsample

    Notes
    -----
    There is some controversy about how to do the fitting. The ML model given
    in [1]_ is based on the assumption that error is roughly proportional to
    magnitude of observation, reasonable for enzyme kinetics but not reasonable
    for rarefaction data. Here we just do a nonlinear curve fit for the
    parameters using least-squares.

    References
    ----------
    .. [1] Raaijmakers, J. G. W. 1987 Statistical analysis of the
       Michaelis-Menten equation. Biometrics 43, 793-803.

    """
    counts = _validate(counts)

    n_indiv = counts.sum()
    if params_guess is None:
        S_max_guess = observed_otus(counts)
        B_guess = int(round(n_indiv / 2))
        params_guess = (S_max_guess, B_guess)

    # observed # of OTUs vs # of individuals sampled, S vs n
    xvals = np.arange(1, n_indiv + 1)
    ymtx = np.empty((num_repeats, len(xvals)), dtype=int)
    for i in range(num_repeats):
        ymtx[i] = np.asarray([observed_otus(subsample(counts, n))
                              for n in xvals], dtype=int)
    yvals = ymtx.mean(0)

    # Vectors of actual vals y and number of individuals n.
    def errfn(p, n, y):
        return (((p[0] * n / (p[1] + n)) - y) ** 2).sum()

    # Return S_max.
    return fmin_powell(errfn, params_guess, ftol=1e-5, args=(xvals, yvals),
                       disp=False)[0]


def observed_otus(counts):
    """Calculate the number of distinct OTUs.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    int
        Distinct OTU count.

    """
    counts = _validate(counts)
    return (counts != 0).sum()


def osd(counts):
    """Calculate observed OTUs, singles, and doubles.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    osd : tuple
        Observed OTUs, singles, and doubles.

    See Also
    --------
    observed_otus
    singles
    doubles

    Notes
    -----
    This is a convenience function used by many of the other measures that rely
    on these three measures.

    """
    counts = _validate(counts)
    return observed_otus(counts), singles(counts), doubles(counts)


def robbins(counts):
    """Calculate Robbins' estimator for the probability of unobserved outcomes.

    Robbins' estimator is defined as

    .. math::

       \\frac{F_1}{n+1}

    where :math:`F_1` is the number of singleton OTUs.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        Robbins' estimate.

    Notes
    -----
    Robbins' estimator is defined in [1]_. The estimate computed here is for
    :math:`n-1` counts, i.e. the x-axis is off by 1.

    References
    ----------
    .. [1] Robbins, H. E (1968). Ann. of Stats. Vol 36, pp. 256-257.

    """
    counts = _validate(counts)
    return singles(counts) / counts.sum()


def shannon(counts, base=2):
    """Calculate Shannon entropy of counts (H), default in bits.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.
    base : scalar, optional
        Logarithm base to use in the calculations.

    Returns
    -------
    double
        Shannon diversity index H.

    Notes
    -----
    The implementation here is based on the description given in the SDR-IV
    online manual [1]_, except that the default logarithm base used here is 2
    instead of :math:`e`.

    References
    ----------
    .. [1] http://www.pisces-conservation.com/sdrhelp/index.html

    """
    counts = _validate(counts)
    freqs = counts / counts.sum()
    nonzero_freqs = freqs[freqs.nonzero()]
    return -(nonzero_freqs * np.log(nonzero_freqs)).sum() / np.log(base)


def simpson(counts):
    """Calculate Simpson's index.

    Simpson's index is defined as 1 - dominance.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        Simpson's index.

    See Also
    --------
    dominance

    Notes
    -----
    The implementation here is ``1 - dominance`` as described in [1]_. Other
    references (such as [2]_) define Simpson's index as ``1 / dominance``.

    References
    ----------
    .. [1] http://folk.uio.no/ohammer/past/diversity.html
    .. [2] http://www.pisces-conservation.com/sdrhelp/index.html

    """
    counts = _validate(counts)
    return 1 - dominance(counts)


def simpson_e(counts):
    """Calculate Simpson's evenness measure E.

    Simpson's E is defined as

    .. math::

       E=\\frac{1 / D}{S_{obs}}

    where :math:`D` is dominance and :math:`S_{obs}` is the number of observed
    OTUs.

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        Simpson's evenness measure E.

    See Also
    --------
    dominance
    enspie
    simpson

    Notes
    -----
    The implementation here is based on the description given in [1]_.

    References
    ----------
    .. [1] http://www.tiem.utk.edu/~gross/bioed/bealsmodules/simpsonDI.html

    """
    counts = _validate(counts)
    return enspie(counts) / observed_otus(counts)


def singles(counts):
    """Calculate number of single occurrences (singletons).

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    int
        Singleton count.

    """
    counts = _validate(counts)
    return (counts == 1).sum()


def strong(counts):
    """Calculate Strong's dominance index (Dw).

    Parameters
    ----------
    counts : 1-D array_like, int
        Vector of counts.

    Returns
    -------
    double
        Strong's dominance index (Dw).

    Notes
    -----
    Strong's dominance index is defined in [1]_. The implementation here is
    based on the description given in the SDR-IV online manual [2]_.

    References
    ----------
    .. [1] Strong, W. L., 2002 Assessing species abundance uneveness within and
       between plant communities. Community Ecology, 3, 237-246.
    .. [2] http://www.pisces-conservation.com/sdrhelp/index.html

    """
    counts = _validate(counts)
    n = counts.sum()
    s = observed_otus(counts)
    i = np.arange(1, len(counts) + 1)
    sorted_sum = np.sort(counts)[::-1].cumsum()
    return (sorted_sum / n - (i / s)).max()