Normalize feature coverages according to spike-in controls with a known ratio

Given a dataframe of features and their coverages, this function will normalize the coverages of the features according to the coverages of the spike-in controls. Both the control and experimental features should be mixed into the same dataframe, and the control features will be identified based on internal data.

Usage

jgi_normalize_features(features, samples, method = "lm")

Arguments

features

A dataframe with feature_id and coverage columns

samples

A dataframe with sample_id, MIX, and sequins_pg columns

method

The regression engine to use. Choices are "lm" or "glm"

Value

A tibble with nested dataframes for each sample, containing the original coverage data, the normalized coverage data, and the model information

Details

Using the control features and their known stoichiometry, a regression model is built converting the coverage to a picogram (pg) amount of DNA. This model is run against the experimental data to convert their coverages into pg amounts of DNA.

This function requires the samples dataframe because it contains the MIX and sequins_pg columns, which are needed to normalize the features.

The output is a sort of intermediate file where a user can get the normalized experimental data with the get_normalized_features() function, which is the input to the main qSIP2 workflow. Or, the user can get the normalized control data with the get_normalized_controls() function in order to see how well the model fits the spike-in controls, and to assess how well each individual spike-in sequences is behaving.

See also

Other "spike-ins": fit_regression_model(), get_normalized_controls(), get_normalized_features(), jgi_feature_df(), jgi_sample_df(), jgi_source_df()