Ninetails requires the following input data to operate:

• POD5 files - for the signal data extraction
• Dorado summary file - for tail lengths estimation, poly(A) coordinates & read/file bindings. Must contain columns: read_id, filename, poly_tail_length, poly_tail_start, poly_tail_end

Warning

The cDNA pipeline is currently under construction. Please do not use it for production analyses.

The cDNA pipeline requires:

• POD5 files - for the signal data extraction
• BAM file - aligned cDNA reads with basecalled sequences (required for read orientation classification)
• Dorado summary file - for tail lengths estimation and poly(A) coordinates

The cDNA pipeline classifies reads as polyA, polyT, or unidentified using Dorado-style SSP/VNP primer matching before processing.

Warning

Current pre-release versions of the package work with Guppy basecaller 6.0.0 and lower. This pipeline is maintained for backward compatibility only and will not be further optimised. For new analyses, use the Dorado DRS pipeline.

Ninetails requires the following input data to operate in legacy mode:

• multifast5 files basecalled by Guppy - for the signal data extraction
• sequencing_summary.txt file - for file ID extraction
• an output of Nanopolish polya function (tsv file) - to obtain the tail segmentation data*

Therefore, please make sure that the third-party software necessary for the steps preceding the use of Ninetails is installed (Nanopolish, Guppy) and/or that you have all the required input files.

For neural network component of Ninetails (data classification), Python is also required.

*since version 1.0.2, Ninetails is also compatible with tailfindR. However, we still strongly recommend to use it with nanopolish.