neuroharmony.ComBat

class neuroharmony.ComBat(features, covariates, eliminate_variance)

ComBat model for harmonization.

A wrapper to the NeuroCombat harmonization as described by Fortin et al., (2018).

Parameters

featureslist

List of features to be harmonized with ComBat.

covariates

List of covariates for which the variance needs to be conserved.

eliminate_variance

List of variables for which the variance needs to be eliminated.

Attributes

harmonized_NDFrame of shape [n_subjects, n_features]

ComBat harmonized DataFrame.

fit(df)

Fit the model.

Fit all the transforms one after the other and transform the data, then fit the transformed data using the final estimator.

Parameters

df: NDFrame of shape [n_subjects, n_features]

Training data. Must fulfil input requirements of first step of the pipeline.

Returns

selfComBat

This estimator

fit_transform(df)

Fit to data, then transform it.

Fits transformer to df and y with optional parameters fit_params and returns a transformed version of df.

Parameters

df: NDFrame of shape [n_subjects, n_features]

Training set.

Returns

harmonized_: NDFrame of shape [n_samples, n_features_new]

Transformed array.

transform(df, y=None)

Run ComBat normalization.

Applies trained model.

Parameters

df: NDFrame of shape [n_subjects, n_features]

Dataframe with data for each subject. The dataframe has to contain the features to be harmonized pandas the covariates which you want to use to harmonization.

Returns

harmonized: NDFrame of shape [n_subjects, n_features]

Dataframe with the harmonized data.

Raises

ValueError:

If there are missing features among the covariates or the features.

Examples

>>> ixi = DataSet('data/raw/IXI').data
>>> features = ['Left-Lateral-Ventricle', 'Left-Inf-Lat-Vent', ]
>>> covariates = ['Gender', 'Dataset', 'Age']
>>> combat = ComBat(features, covariates)
>>> harmonized = combat(ixi)
>>> harmonized

subject_id	Left-Lateral-Ventricle	Left-Inf-Lat-Vent	Gender	Dataset	Age
sub-002-00-2	0.007666	0.000134	0	2	31
sub-012-00-1	0.003778	0.000064	1	1	14
sub-013-00-1	0.012818	0.000464	1	1	17
sub-014-00-1	0.006289	0.000167	0	1	19
sub-015-00-1	0.003310	0.000154	1	1	26