Skip to contents

qSIP feature table class

Source: R/qsip_objects.R
qsip_feature_data.Rd

The qsip_feature_data object holds validated feature metadata.

Usage

qsip_feature_data(data, feature_id = "feature_id", type = "counts")

Arguments

data

(dataframe) ASV/OTU table or equivalent

feature_id

(string) Column name with unique taxa IDs

type

(string, default: counts) The type of numerical data, either counts, coverage, normalized or relative

Value

A validated object of the qsip_feature_data type

Details

qsip_feature_data() is not a typical function, but rather a class constructor that instantiates a new qsip_feature_data object. The constructor takes a data.frame as input and returns a validated qsip_feature_data object.

The qsip_feature_data object is used to hold feature metadata, such as amplicon sequence variants (ASVs), operational taxonomic units (OTUs), metagenome-assembled genomes (MAGs), etc.

The data argument takes a data.frame that has the feature IDs as a column designated with the feature_id argument. Each row corresponds to a unique feature (amplicon, MAG, etc) and each subsequent row corresponds to a unique sample.

The type argument is used to designate the type of data in the data argument. It should most likely be counts for amplicon data, and coverage for metagenome data (including normalizations like TPM). If the data is relative abundances, the type argument should be set to relative. Overall, the choice won't much affect the results from the qSIP analysis, but choosing an accurate type will help with the validation checks.

Internally, qsip_feature_data renames the metadata columns to be standardized to MISIP terminology. A data.frame with the standardized names can be extracted back out of the object using the get_dataframe() method, and the optional original_headers argument can be set to TRUE to return the original column names.

There are several validation checks run on the data on the data.frame:

  • The data argument must contain a data.frame, including a tibble

  • The feature_id argument must be a column name in the data.frame

  • The feature_id column must contain unique values per row

  • The type argument must be one of counts, coverage or relative

    • The type argument is counts by default, and in this case the values in the data argument must be integers

    • If type is set to relative the values in the data argument must be numeric and the values must sum to 1 for each row

    • If type is set to coverage, the values in the data argument must be numeric

    • If the type is set to normalized then the values are assumed to be pre-normalized and additional transformations will not be done.

  • All values in the data argument must be non-negative

See also

Other "qSIP Objects": qsip_data(), qsip_sample_data(), qsip_source_data()