linkMultipleVariants: Process an experiment with multiple variable sequences
In fmicompbio/mutscan: Preprocessing and Analysis of Deep Mutational Scanning Data

linkMultipleVariants

R Documentation

Process an experiment with multiple variable sequences

Description

This function enables the processing of data sets with multiple variable sequences, which should potentially be handled in different ways. For example, a barcode association experiment with two variable sequences (the barcode and the biological variant) that need to be processed differently, e.g. in terms of matching to wildtype sequences or collapsing of similar sequences. In contrast, while digestFastqs allow the specification of multiple variable sequences (within each of the forward and reverse reads), they will be concatenated and processed as a single unit.

Usage

linkMultipleVariants(combinedDigestParams = list(), ...)

Arguments

`combinedDigestParams`	A named list of arguments to `digestFastqs` for the combined ("naive") run.
`...`	Additional arguments providing arguments to `digestFastqs` for the separate runs (processing each variable sequence in turn). Each argument must be a named list of arguments to `digestFastqs`. In addition, arguments `collapseMaxDist`, `collapseMinScore` and `collapseMinRatio` can be specified, and will be passed on to `collapseMutantsBySimilarity`.

Details

linkMultipleVariants will process the input in the following way:

First, run digestFastqs with the parameters provided in combinedDigestParams. Typically, this will be a "naive" counting run, where the frequencies of all observed variants are tabulated. The variable sequences within the forward and reverse reads, respectively, will be processed as a single sequence.
Next, run digestFastqs with each of the additional parameter sets provided (...). Each of these should correspond to a single variable sequence from the combined run (i.e., if there are two Vs in the element specifications in the combined run, there should be two additional parameter sets provided, each corresponding to the processing of one variable sequence part). It is assumed that the order of the additional arguments correspond to the order of the variable sequences in the combined run, in such a way that if the variable sequences extracted in each of the separate runs are concatenated in the order that the parameter sets are provided to linkMultipleVariants, they will form the variable sequence extracted in the combined run.
The result of each of the separate runs is a 'conversion table', containing the final set of identified sequence variants as well as all individual sequences corresponding to each of them. This is then combined with the count table from the combined, "naive" run in order to create an aggregated count table. More precisely, each sequence in the combined run is split into the constituent variable sequences, and each variable sequence is then matched to the output from the right separate run, from which the final feature ID (mutant name, or collapsed sequence) will be extracted and used to replace the original sequence in the combined count table. Once all the matches are done, rows with NAs (where no match could be found in the separate run) are removed and the counts are aggregated across all identical combinations of variable sequences.

In order to define the elementsForward and elementsReverse arguments for the separate runs, a strategy that often works is to simply copy the arguments from the combined run, and successively replace all but one of the 'V's by 'S'. This will effectively process one variable sequence at the time, while keeping all other elements of the reads consistent (since this can affect e.g. filtering criteria). Note that to process individual variable sequences in the reverse read, you also need to swap the 'forward' and 'reverse' specifications (since digestFastqs requires a forward read).

Value

A list with the following elements:

countAggregated - a tibble with columns corresponding to each of the variable sequences, and a column with the total observed read count for the combination.
convSeparate - a list of conversion tables from the respective separate runs.
outCombined - the digestFastqs output for the combined run.