Merges ninetails tabular outputs (read classes and nonadenosine residue data) to produce one concise table.

Combines the read_classes and nonadenosine_residues data frames into a single wide-format tibble with one row per read. The prediction column from the residue data is spread into three columns (prediction_C, prediction_G, prediction_U) via spread_nonA_residues, and a CIGAR-like nonA_residues summary column is appended.

Usage

merge_nonA_tables(class_data, residue_data, pass_only = TRUE)

Arguments

class_data

Data frame or tibble containing read_classes predictions produced by the ninetails pipeline. The qc_tag column may be character (Guppy/nanopolish: "PASS", "SUFFCLIP", etc.) or numeric (Dorado: MAPQ score).

residue_data

Data frame or tibble containing non-A residue predictions produced by the ninetails pipeline. The qc_tag column type must match that of class_data.

pass_only

Logical [TRUE]. Applies only when qc_tag is character (Guppy/nanopolish pipeline). If TRUE, only reads tagged as "PASS" are included. If FALSE, reads tagged as "PASS" or "SUFFCLIP" are included. Ignored when qc_tag is numeric (Dorado pipeline), in which case MAPQ > 0 filtering is applied instead.

Value

A tibble with summarised information from both ninetails outputs: all columns from class_data plus prediction_C, prediction_G, prediction_U, and nonA_residues.

Details

Unclassified reads are excluded before merging. The quality-tag filter behaviour depends on the pipeline that produced the input data:

Guppy/nanopolish pipeline (character qc_tag): The pass_only parameter controls filtering. When TRUE, only reads tagged as "PASS" are retained. When FALSE, reads tagged as "PASS" or "SUFFCLIP" are retained.

Dorado pipeline (numeric qc_tag): The qc_tag column contains MAPQ scores. Filtering is based on mapping quality: only reads with qc_tag > 0 are retained. The pass_only parameter is ignored for numeric qc_tag.

After the full join, any NA values in numeric columns are replaced with 0.

Pipeline Compatibility

This function automatically detects whether data originates from the legacy Guppy/nanopolish pipeline (character qc_tag) or the Dorado pipeline (numeric qc_tag) and applies appropriate filtering logic. Both class_data and residue_data must originate from the same pipeline to ensure consistent qc_tag types during the merge operation.

See also

spread_nonA_residues for the reshaping step, summarize_nonA for transcript-level summaries from the merged table, calculate_fisher for statistical testing on the merged table, read_class_single and read_residue_single for loading the input data.

Examples

if (FALSE) { # \dontrun{

# Guppy/nanopolish data (character qc_tag)
merged_tables <- ninetails::merge_nonA_tables(
  class_data = class_data,
  residue_data = residue_data,
  pass_only = TRUE)

# Dorado data (numeric qc_tag) - pass_only is ignored
merged_tables <- ninetails::merge_nonA_tables(
  class_data = dorado_class_data,
  residue_data = dorado_residue_data,
  pass_only = TRUE)

} # }