epistasis.models package

See subpackages for specific models.

epistasis.models.base module

class epistasis.models.base.AbstractModel

Bases: abc.ABC

Abstract Base Class for all epistasis models.

This class sets all docstrings not given in subclasses.

static __new__(self, *args, **kwargs)

Replace the docstrings of a subclass with docstrings in this base class.

add_X(X=None, key=None)

Add X to Xbuilt

Keyword arguments for X:

  • None :
    Uses gpm.binary to construct X. If genotypes are missing they will not be included in fit. At the end of fitting, an epistasis map attribute is attached to the model class.
Parameters:
  • X – see above for details.
  • key (str) – name for storing the matrix.
Returns:

Xbuilt – newly built 2d array matrix

Return type:

numpy.ndarray

add_gpm(gpm)

Add a GenotypePhenotypeMap object to the epistasis model.

fit(X=None, y=None, **kwargs)

Fit model to data.

Parameters:
  • X (None, ndarray, or list of genotypes. (default=None)) – data used to construct X matrix that maps genotypes to model coefficients. If None, the model uses genotypes in the attached genotype-phenotype map. If a list of strings, the strings are genotypes that will be converted to an X matrix. If ndarray, the function assumes X is the X matrix used by the epistasis model.
  • y (None or ndarray (default=None)) – array of phenotypes. If None, the phenotypes in the attached genotype-phenotype map is used.
Returns:

The model is returned. Allows chaining methods.

Return type:

self

fit_transform(X=None, y=None, **kwargs)

Fit model to data and transform output according to model.

Parameters:
  • X (None, ndarray, or list of genotypes. (default=None)) – data used to construct X matrix that maps genotypes to model coefficients. If None, the model uses genotypes in the attached genotype-phenotype map. If a list of strings, the strings are genotypes that will be converted to an X matrix. If ndarray, the function assumes X is the X matrix used by the epistasis model.
  • y (None or ndarray (default=None)) – array of phenotypes. If None, the phenotypes in the attached genotype-phenotype map is used.
Returns:

gpm – The genotype-phenotype map object with transformed genotypes.

Return type:

GenotypePhenotypeMap

gpm

Data stored in a GenotypePhenotypeMap object.

hypothesis(X=None, thetas=None)

Compute phenotypes from given model parameters.

Parameters:
  • X (None, ndarray, or list of genotypes. (default=None)) – data used to construct X matrix that maps genotypes to model coefficients. If None, the model uses genotypes in the attached genotype-phenotype map. If a list of strings, the strings are genotypes that will be converted to an X matrix. If ndarray, the function assumes X is the X matrix used by the epistasis model.
  • thetas (ndarray) – array of model parameters. See thetas property for specifics.
Returns:

y – array of phenotypes predicted by model parameters.

Return type:

ndarray

hypothesis_transform(X=None, y=None, thetas=None)

Transform phenotypes with given model parameters.

Parameters:
  • X (None, ndarray, or list of genotypes. (default=None)) – data used to construct X matrix that maps genotypes to model coefficients. If None, the model uses genotypes in the attached genotype-phenotype map. If a list of strings, the strings are genotypes that will be converted to an X matrix. If ndarray, the function assumes X is the X matrix used by the epistasis model.
  • y (ndarray) – An array of phenotypes to transform.
  • thetas (ndarray) – array of model parameters. See thetas property for specifics.
Returns:

y – array of phenotypes predicted by model parameters.

Return type:

ndarray

lnlike_of_data(X=None, y=None, yerr=None, thetas=None)

Compute the individUal log-likelihoods for each datapoint from a set of model parameters.

Parameters:
  • X (None, ndarray, or list of genotypes. (default=None)) – data used to construct X matrix that maps genotypes to model coefficients. If None, the model uses genotypes in the attached genotype-phenotype map. If a list of strings, the strings are genotypes that will be converted to an X matrix. If ndarray, the function assumes X is the X matrix used by the epistasis model.
  • y (ndarray) – An array of phenotypes to transform.
  • yerr (ndarray) – An array of the measured phenotypes standard deviations.
  • thetas (ndarray) – array of model parameters. See thetas property for specifics.
Returns:

y – array of phenotypes predicted by model parameters.

Return type:

ndarray

lnlike_transform(X=None, y=None, yerr=None, lnprior=None, thetas=None)

Compute the individual log-likelihoods for each datapoint from a set of model parameters and a prior.

Parameters:
  • X (None, ndarray, or list of genotypes. (default=None)) – data used to construct X matrix that maps genotypes to model coefficients. If None, the model uses genotypes in the attached genotype-phenotype map. If a list of strings, the strings are genotypes that will be converted to an X matrix. If ndarray, the function assumes X is the X matrix used by the epistasis model.
  • y (ndarray) – An array of phenotypes to transform.
  • yerr (ndarray) – An array of the measured phenotypes standard deviations.
  • lnprior (ndarray) – An array of priors for a given datapoint.
  • thetas (ndarray) – array of model parameters. See thetas property for specifics.
Returns:

y – array of phenotypes predicted by model parameters.

Return type:

ndarray

lnlikelihood(X=None, y=None, yerr=None, thetas=None)

Compute the individal log-likelihoods for each datapoint from a set of model parameters.

Parameters:
  • X (None, ndarray, or list of genotypes. (default=None)) – data used to construct X matrix that maps genotypes to model coefficients. If None, the model uses genotypes in the attached genotype-phenotype map. If a list of strings, the strings are genotypes that will be converted to an X matrix. If ndarray, the function assumes X is the X matrix used by the epistasis model.
  • y (ndarray) – An array of phenotypes to transform.
  • yerr (ndarray) – An array of the measured phenotypes standard deviations.
  • thetas (ndarray) – array of model parameters. See thetas property for specifics.
Returns:

lnlike – log-likelihood of the model parameters.

Return type:

float

num_of_params

Number of parameters in model.

predict(X=None)

Use model to predict phenotypes for a given list of genotypes.

Parameters:X (None, ndarray, or list of genotypes. (default=None)) – data used to construct X matrix that maps genotypes to model coefficients. If None, the model uses genotypes in the attached genotype-phenotype map. If a list of strings, the strings are genotypes that will be converted to an X matrix. If ndarray, the function assumes X is the X matrix used by the epistasis model.
Returns:y – array of phenotypes.
Return type:ndarray
predict_transform(X=None, y=None, **kwargs)

Transform a set of phenotypes according to the model.

Parameters:
  • X (None, ndarray, or list of genotypes. (default=None)) – data used to construct X matrix that maps genotypes to model coefficients. If None, the model uses genotypes in the attached genotype-phenotype map. If a list of strings, the strings are genotypes that will be converted to an X matrix. If ndarray, the function assumes X is the X matrix used by the epistasis model.
  • y (ndarray) – An array of phenotypes to transform.
Returns:

y_transform – array of phenotypes.

Return type:

ndarray

class epistasis.models.base.BaseModel

Bases: epistasis.models.base.AbstractModel, sklearn.base.RegressorMixin, sklearn.base.BaseEstimator

Base model for defining an epistasis model.

exception epistasis.models.base.SubclassException

Bases: Exception

Subclass Exception for parent classes.

epistasis.models.base.use_sklearn(sklearn_class)

Swap out last class in the inherited stack (Assuming its the BaseModel) with the AbstractModel below. Then, sandwiches the Sklearn class with all other base classes first, followed by the Sklearn class and the AbstractModel.

epistasis.models.classifiers module

epistasis.models.ensemble module

epistasis.models.pipeline module

epistasis.models.utils module

exception epistasis.models.utils.FittingError

Bases: Exception

Exception Subclass for X matrix errors.

exception epistasis.models.utils.XMatrixException

Bases: Exception

Exception Subclass for X matrix errors.

epistasis.models.utils.arghandler(method)

Points methods to argument handlers. Assumes each argument has a corresponding method attached to the object named “_{argument}”. These methods given default values to arguments.

Ignores self and kwargs

Module contents