Is there a way to transform both the data matrix and labels in a single pipeline before fitting a model in sklearn?

Question

Suppose I have

• a data matrix X (numpy ndarray, all features are numeric)
• labels y (numpy array of string).

I want to apply SimpleImputer(strategy='mean') and StandardScaler() to X, and OrdinalEncoder() to y. After receiving the transformed data, I want to use LogisticRegression() estimator for RFE() to select 2 most important features from the data.
Is there a nice way to create a single Pipeline() to perform this task?
If not then how can this be done?

I want to use the Pipeline() instance like this :

pipe = Pipeline(<some code>)
important_matrix = pipe.fit_transform(X, y)