Additive Log-Ratio Normalization

This function performs Additive Log-Ratio (ALR) transformation on compositional data. ALR transforms each component by taking the logarithm of the ratio of the component to a reference component. This is useful for analyzing compositional data where the relative proportions are of interest.

Usage

normalize_alr(x)

# S3 method for class 'glyexp_experiment'
normalize_alr(x)

# S3 method for class 'matrix'
normalize_alr(x)

# Default S3 method
normalize_alr(x)

Arguments

x

Either a glyexp_experiment object or a matrix. If a matrix, rows should be variables and columns should be samples.

Value

Returns the same type as the input. If x is a glyexp_experiment, returns a glyexp_experiment with an ALR-transformed expression matrix. If x is a matrix, returns an ALR-transformed matrix. In both cases, the returned matrix has the same number of rows (variables) as the input: the automatically selected reference variable is retained as a row of zeros after transformation, and all other variables contain log-ratios relative to this reference. The reference variable is the geometric-median variable described above, and its row position/name is preserved. Note that the resulting values are on the log scale and can be negative.

Details

The formula is: alr(x)_i = log(x_i / x_ref) where x_ref is the reference component.

This function requires all values to be strictly positive (no zeros or NA values). Users must handle zeros and missing values before calling this function.

The reference variable is automatically selected as the geometric median (the variable with the smallest sum of log-ratio distances to all other variables), which provides the most stable reference.

This function is a wrapper around compositions::alr().