mimic.model_infer package

Submodules

mimic.model_infer.cLV module

class mimic.model_infer.cLV.CompositionalLotkaVolterra(P=None, T=None, U=None, denom=None, pseudo_count=0.001)

Bases: object

Inference for compositional Lotka-Volterra.

\[\]

rac{d}{dt} log( rac{pi_i(t)}{pi_D(t)}) = g_i + sum^D_{j=1} (A_{ij} pi_j(t)) + sum^P_{p=1} (B_{ip} u_p(t)

Parameters

P : A list of T_x by D dimensional numpy arrays of estimated relative abundances. T : A list of T_x by 1 dimensional numpy arrays giving the times of each observation x. U : An optional list of T_x by P numpy arrays of external perturbations for each x. denom : integer, id for taxa in denominator of log ratio pseudo_count : float, default=1e-3, a small number to replace zero counts

get_params() → tuple[ndarray, ndarray, ndarray]

get_regularizers() → Tuple[float | None, float | None, float | None, float | None]

predict(p0, times, u=None) → ndarray

Predict relative abundances from initial conditions.

Parameters

p0 : the initial observation, a D-dim numpy array times : a T by 1 numpy array of sample times u : a T by P numpy array of external perturbations

Returns

y_preda T by D numpy array of predicted relative

abundances. Since we cannot predict initial conditions, the first entry is set to the array of -1.

set_regularizers(alpha, r_A, r_g, r_B)

train(verbose=False, folds=10) → None

Estimate regularization parameters and CLV model parameters.

mimic.model_infer.cLV.choose_denom(P) → int

Pick a denominator for additive log-ratio transformation.

mimic.model_infer.cLV.compute_prediction_error(X, P, U, T, A, g, B, denom_ids) → float

mimic.model_infer.cLV.compute_rel_abun(x, denom) → ndarray

mimic.model_infer.cLV.construct_alr(P, denom, pseudo_count=0.001) → list[ndarray]

Compute the additive log ratio transformation with a given choice of denominator. Assumes zeros have been replaced with nonzero values.

mimic.model_infer.cLV.elastic_net_clv(X, P, U, T, Q_inv, alpha, r_A, r_g, r_B, tol=0.001, verbose=False, max_iter=10000) → tuple[ndarray, ndarray, ndarray]

mimic.model_infer.cLV.estimate_elastic_net_regularizers_cv(X, P, U, T, denom, folds, no_effects=False, verbose=False) → Tuple[float, float, float, float] | None

mimic.model_infer.cLV.estimate_relative_abundances(Y, pseudo_count=0.001) → list[ndarray]

Adds pseudo counts to avoid zeros and compute relative abundances

Parameters

Ya list sequencing counts or observed concentrations

per sequence

pseudo_countpseudo count, specific in relation to

the relative proportions of each taxon.

Returns

P : relative abundances with pseudo counts

mimic.model_infer.cLV.estimate_ridge_regularizers_cv(X, P, U, T, denom, folds, no_effects=False, verbose=False) → tuple[float, float, float] | None

mimic.model_infer.cLV.predict(x, p, u, times, A, g, B, denom) → ndarray

Make predictions from initial conditions

mimic.model_infer.cLV.ridge_regression_clv(X, P, U, T, r_A=0, r_g=0, r_B=0) → tuple[ndarray, ndarray, ndarray]

Computes estimates of A, g, and B using least squares.

Parameters

X : a list of T x yDim-1 numpy arrays U : a list of T x uDim numpy arrays T : a list of T x 1 numpy arrays with the time of each observation

Returns

mimic.model_infer.infer_VAR_bayes module

class mimic.model_infer.infer_VAR_bayes.infer_VAR(data=None, coefficients=None, intercepts=None, covariance_matrix=None, dataS=None)

Bases: BaseInfer

infer_VAR class for performing inference on VAR models.

This class provides methods for importing data, running inference on the VAR model, and performing posterior sampling and analysis.

Args:

data (numpy.ndarray): The data to perform inference on.

Methods:

run_inference(method=’default’, **kwargs):

Runs the inference process for the VAR model.

Returns:

None

make_plot_stacked(dataX, dataS)

Creates a stacked plot of abundance and metabolite data.

Args: dataX (numpy.ndarray): The abundance data. dataS (numpy.ndarray): The metabolite data.

Returns: None

plot_heatmap(idata, matrices=None, true_values=None)

Plots heatmaps of the inferred matrices.

Args: idata (arviz.InferenceData): The inference data. matrices (list of str): List of keys for matrices in idata.posterior to plot. true_values (list of numpy.ndarray): List of true matrices to compare with.

Returns: None

posterior_analysis(data_filename=None, netcdf_filename=None, A=None, B=None)

Performs posterior analysis and visualizes the results.

Args: data_filename (str, optional): The filename for the .npz data file. netcdf_filename (str, optional): The filename for the NetCDF file with inference results. A (numpy.ndarray, optional): The true matrix A to compare against. B (numpy.ndarray, optional): The true matrix B to compare against.

Returns: None

run_inference(method='default', **kwargs)

Runs the inference process for the VAR model.

Args:

method (str): Specifies which inference method to run.

Options are ‘default’, ‘large’, ‘xs’, ‘large_xs’. Default is ‘default’.

**kwargs: Additional keyword arguments to pass to the inference methods.

Returns:

None

set_parameters(data: ndarray | DataFrame | list | tuple | None = None, coefficients: ndarray | list | None = None, intercepts: ndarray | list | None = None, covariance_matrix: ndarray | list | None = None, dataS: ndarray | DataFrame | list | tuple | None = None, priors: Dict[str, Any] | None = None) → None

Sets the parameters for the infer_VAR instance.

Allows optional specification of all model parameters. Parameters not provided (None) are left unchanged.

Parameters:

data (Optional[Union[np.ndarray, pd.DataFrame, list, tuple]]): The data to perform inference on. coefficients (Optional[Union[np.ndarray, list]]): Coefficients of the VAR model. intercepts (Optional[Union[np.ndarray, list]]): Intercepts of the VAR model. covariance_matrix (Optional[Union[np.ndarray, list]]): Covariance matrix of the VAR model. dataS (Optional[Union[np.ndarray, pd.DataFrame, list, tuple]]): The secondary data (e.g., metabolite data). priors (Optional[Dict[str, Any]]): A dictionary of prior distributions.

mimic.model_infer.infer_gMLV_lm module

class mimic.model_infer.infer_gMLV_lm.Ridge1(alphas=None, num_species=3)

Bases: BaseEstimator, RegressorMixin

Custom ridge regression class

fit(X, y) → None

get_params(deep=True) → dict

Get parameters for this estimator.

Parameters

deepbool, default=True

If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns

paramsdict

Parameter names mapped to their values.

predict(X) → ndarray

set_score_request(*, sample_weight: bool | None | str = '$UNCHANGED$') → Ridge1

Request metadata passed to the score method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to score if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to score.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters

sample_weightstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for sample_weight parameter in score.

Returns

selfobject

The updated object.

class mimic.model_infer.infer_gMLV_lm.Ridge2(alphas=None, num_species=3, num_pert=1)

Bases: BaseEstimator, RegressorMixin

Custom ridge regression class

fit(X, y) → None

get_params(deep=True) → dict

Get parameters for this estimator.

Parameters

deepbool, default=True

If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns

paramsdict

Parameter names mapped to their values.

predict(X) → ndarray

set_score_request(*, sample_weight: bool | None | str = '$UNCHANGED$') → Ridge2

Request metadata passed to the score method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to score if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to score.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters

sample_weightstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for sample_weight parameter in score.

Returns

selfobject

The updated object.

mimic.model_infer.infer_gMLV_lm.do_bootstrapping(X, F, num_species, a0, a1, nt, nboots=100) → None

mimic.model_infer.infer_gMLV_lm.do_final_fit_Ridge1(X, F, num_species, a0, a1) → tuple[list[float], list[list[float]]]

mimic.model_infer.infer_gMLV_lm.do_final_fit_Ridge2(X, F, num_species, num_pert, a0, a1, a2) → tuple[list[float], list[list[float]], list[list[float]]]

mimic.model_infer.infer_gMLV_lm.fit_alpha_Ridge1(X, F, num_species, n_a0, n_a1) → tuple[float, float]

mimic.model_infer.infer_gMLV_lm.fit_alpha_Ridge2(X, F, num_species, num_pert, n_a0, n_a1, n_a2) → tuple[float, float, float]

mimic.model_infer.infer_gMLV_lm.fit_alpha_lasso(X, S, n_a) → tuple[float, float]

mimic.model_infer.infer_gMLV_lm.linearise_time_course_metabolites(sobs, yobs, times) → tuple[ndarray, ndarray]

mimic.model_infer.infer_gMLV_lm.linearize_time_course(yobs, times) → tuple[ndarray, ndarray]

mimic.model_infer.infer_gMLV_lm.linearize_time_course_16S(yobs, times) → tuple[ndarray, ndarray]

mimic.model_infer.infer_gMLV_lm.linearize_time_course_16S_u(yobs, times, u) → tuple[ndarray, ndarray]

mimic.model_infer.infer_gMLV_lm.plot_alpha_lasso(X, S, n_a) → None

mimic.model_infer.infer_gMLV_lm.plot_coeffs(tX, tF) → None

mimic.model_infer.infer_gMLV_lm.ridge_fit(X, F, alphas, num_species) → ndarray

mimic.model_infer.infer_gMLV_lm.ridge_fit_pert(X, F, alphas, num_species, num_pert) → ndarray

mimic.model_infer.infer_gMLV_lm.ridge_fit_test(tX, tF, num_species, cRidge=<class 'mimic.model_infer.infer_gMLV_lm.Ridge1'>) → None

mimic.model_infer.multi_penalty_lasso module

class mimic.model_infer.multi_penalty_lasso.MultiPenaltyLasso(alpha)

Bases: BaseEstimator, RegressorMixin

Linear regression with non-uniform L1 regularizer.

Minimizes objective function:

1 / (2 * n_samples) * ||y - X * w||^2_2 + ||alpha * w||_1

where:

alpha = diag(alpha_1, alpha_2, ..., alpha_{num_samples})

and all alpha are non-zero

fit(X, y)

get_params(deep=True) → dict

Get parameters for this estimator.

Parameters

deepbool, default=True

If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns

paramsdict

Parameter names mapped to their values.

non_zero_penalties(X, y) → ndarray

penalised_lasso(X, y) → ndarray

predict(X) → ndarray

set_score_request(*, sample_weight: bool | None | str = '$UNCHANGED$') → MultiPenaltyLasso

Request metadata passed to the score method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to score if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to score.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters

sample_weightstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for sample_weight parameter in score.

Returns

selfobject

The updated object.

mimic.model_infer.multi_penalty_lasso.fit_alpha_MPLasso(X, y, n_a) → ndarray

Module contents