Skip to contents

Extracts fragments of poly(A) tail signal (Dorado mode) containing potential modifications along with their delimitation (positional indices; coordinates) within the tail.

Source: R/ninetails_core_functions_dorado_DRS.R
split_tail_centered_dorado.Rd

This function processes raw poly(A) tail signal and Dorado-derived pseudomoves to identify and extract signal segments (chunks) potentially corresponding to modified positions (e.g., non-A residues). Each extracted chunk spans 100 signal points, centered on the midpoint of a pseudomove run.

Usage

split_tail_centered_dorado(readname, tail_feature_list)

Arguments

readname

character string. Name of the given read within the analyzed dataset.

tail_feature_list

list object produced by create_tail_features_list_dorado Dorado-tail feature extraction function. Must contain $tail_signal (numeric vector) and $tail_pseudomoves (integer vector) for each read.

Value

a nested list where each element corresponds to a signal fragment. Each fragment is itself a list with three entries:

  • chunk_sequence: numeric vector of raw signal values

  • chunk_start_pos: integer, start index of the chunk

  • chunk_end_pos: integer, end index of the chunk

Details

In the Dorado pipeline, moves are not used: * retrieving them from BAM files is computationally expensive * processing is non-intuitive

Instead, only pseudomoves are considered. As a safeguard against Dorado’s tendency to extend poly(A) boundaries into the transcript body, the last 3 pseudomove values are forced to 0. This prevents misclassification of transcript nucleotides as part of the tail.

Candidate modification regions are detected by: * run-length encoding (RLE) of the pseudomove vector * filtering runs of pseudomoves with length ≥ 5

Extracted fragments are padded/imputed if they extend beyond signal boundaries: * upstream/downstream missing values (NAs) are replaced * imputation is based on random draws from the 5 most frequent signal values

The function returns a list object (nested), where each element represents one candidate modification region, containing: * `chunk_sequence`: the raw signal subsequence (length = 100, imputed if needed) * `chunk_start_pos`: starting index of the subsequence * `chunk_end_pos`: ending index of the subsequence

Examples

if (FALSE) { # \dontrun{
ninetails::split_tail_centered_dorado(
  readname = "1234-anexample-r3adn4m3",
  tail_feature_list = tail_feature_list
)
} # }