An R package for finding non-adenosine residues in poly(A) tails of Oxford Nanopore sequencing reads

Ninetails detects and characterises non-adenosine nucleotides embedded within poly(A) tails of Oxford Nanopore sequencing reads. It currently supports three pipelines:

WARNING! cDNA pipeline under construction! Please do not use it for now!

Ninetails can distinguish characteristic signal signatures of four nucleotide types: adenosines (A), cytosines (C), guanosines (G), and uridines (U).

The software is still under active development. All suggestions for improvement are welcome. Please note that the code may change frequently, so use it with caution.

Ninetails has been tested on Linux Mint 20.3, Ubuntu 20.04.3 and Windows 11 with R 4.1.2, R 4.2.0 and R 4.2.1.

📘 Full documentation: https://LRB-IIMCB.github.io/ninetails/

📖 Wiki (tutorials & guides): https://github.com/LRB-IIMCB/ninetails/wiki

Ninetails is not currently available on CRAN or Bioconductor. Install it using devtools:

install.packages("devtools")
devtools::install_github('LRB-IIMCB/ninetails')
library(ninetails)

Note for Windows users

Before installing devtools on Windows, install Rtools so packages compile correctly: https://cran.r-project.org/bin/windows/Rtools/

Installation takes approximately 20 seconds on a typical PC. Additional time is required to install and configure third-party components (Dorado, POD5 Python module, Keras).

Ninetails requires additional components to operate. See the Wiki for full installation instructions.

All pipeline wrappers share the check_tails prefix. Choose the appropriate function for your data type.

The recommended pipeline for direct RNA sequencing (DRS) data basecalled with Dorado ≥ 1.0.0 using POD5 format.

results <- ninetails::check_tails_dorado_DRS(
  dorado_summary = "path/to/dorado_alignment_summary.txt",
  pod5_dir       = "path/to/pod5_dir/",
  num_cores      = 2,
  qc             = TRUE,
  save_dir       = "~/output/",
  prefix         = "experiment1",
  part_size      = 40000,
  cleanup        = FALSE
)

Parameters:

For complementary DNA (cDNA) sequencing data. Extends the DRS pipeline with BAM file processing for sequence extraction and automatic read orientation classification (polyA vs polyT).

results <- ninetails::check_tails_dorado_cDNA(
  bam_file       = "path/to/aligned_cdna.bam",
  dorado_summary = "path/to/dorado_summary.txt",
  pod5_dir       = "path/to/pod5_dir/",
  num_cores      = 2,
  qc             = TRUE,
  save_dir       = "~/output/",
  prefix         = "experiment1",
  part_size      = 40000,
  cleanup        = FALSE
)

Additional parameter vs DRS:

The cDNA pipeline classifies each read as polyA, polyT, or unidentified using Dorado-style SSP/VNP primer matching before processing. Output tables include an additional tail_type column.

For DRS data basecalled with Guppy ≤ 6.0.0 using fast5 format and Nanopolish poly(A) coordinates. This pipeline is no longer actively developed (critical bug fixes only).

Warning

Current pre-release versions of the package work with Guppy basecaller 6.0.0 and lower. Please be aware to use a compatible version of basecaller.

results <- ninetails::check_tails_guppy(
  polya_data = system.file('extdata', 'test_data',
                            'nanopolish_output.tsv',
                            package = 'ninetails'),
  sequencing_summary = system.file('extdata', 'test_data',
                                    'sequencing_summary.txt',
                                    package = 'ninetails'),
  workspace = system.file('extdata', 'test_data',
                           'basecalled_fast5',
                           package = 'ninetails'),
  num_cores      = 2,
  basecall_group = 'Basecall_1D_000',
  pass_only      = TRUE,
  save_dir       = '~/output/')

Before running, ensure that the Nanopolish output file, sequencing summary, and fast5 directory all correspond to the same set of reads. Complete discrepancies will cause an error; partial mismatches will produce a warning and the affected reads will be omitted.

Note

Ninetails does not support single fast5 files. Convert them to multi-fast5 format with ont-fast5-api before using this pipeline.

Legacy classification codes (Guppy pipeline only — these differ from the Dorado pipelines):

All pipelines return a named list with two data frames and save them as tab-delimited text files in save_dir. A log file is also written for each run.

Complete accounting of all reads in the analysis. Each read is assigned one class and one comment code.

Classification codes:

Modification-level detail for decorated reads only. Each row is one predicted non-adenosine residue. A read may have multiple rows if multiple modifications were detected.

• Signal transformations can place a heavy load on memory, especially for full sequencing runs. Adjust part_size to manage memory usage.
• Ninetails relies on external poly(A) boundary detection (Dorado / Nanopolish / tailfindr) and may underestimate terminal modifications at the last 1–2 nucleotides of the tail.
• Since version 1.0.2, Ninetails is compatible with tailfindr output via tailfindr_compatibility().

Please cite Ninetails as:

Gumińska, N., Matylla-Kulińska, K., Krawczyk, P.S. et al. Direct profiling of non-adenosines in poly(A) tails of endogenous and therapeutic mRNAs with Ninetails. Nat Commun 16, 2664 (2025). https://doi.org/10.1038/s41467-025-57787-6

If you encounter a bug, please open an issue on GitHub. To help diagnose the problem, provide a minimal reproducible example (5–10 reads with corresponding Dorado summary / Nanopolish output / sequencing summary), so the error can be reproduced and fixed.

Any issues regarding Ninetails should be addressed to Natalia Gumińska (nguminska (at) iimcb.gov.pl).

Ninetails was developed in the Laboratory of RNA Biology (Dziembowski Lab) at the International Institute of Molecular and Cell Biology in Warsaw.