A Dynamic Programming Approach to Segment ONT Signals. Dynamont is a segmentation/resquiggling tool for ONT signals. Dynamont is currently designed for RNA002 and RNA004 data!

• Currently, I applied the trained transition parameters from the RNA004 model to the DNA R10 models. These should be fine-tuned for the DNA models.

---

• Installation
  • Pypi/pip
  • Conda
• Usage
• Default models:
• Output
  • Example Output
• Exit-Codes

---

Installation

Pypi/pip

pip install dynamont

Conda

conda config --add channels jannessp # to install all dependencies from the correct channel
conda create -n dynamont jannessp::dynamont
conda activate dynamont

Usage

# segment a dataset
dynamont-resquiggle -r <path/to/pod5/dataset/> -b <basecalls.bam> --mode basic -o <output.csv> -p <pore>

# train model
dynamont-train -r <path/to/pod5/dataset/> -b <basecalls.bam> --mode basic -o <output/path> -p <pore>

# choosing a pore will automatically load the default model for that pore, a custom model can be used with the parameter --pore_model <model/path>

Default models:

• rna_r9 (tested)
• rna_rp4 (tested)
• dna_r9 not available yet
• dna_r10.4.1 260 bps (not tested)
• dna_r10.4.1 400 bps (not tested)

Output

Dynamont produces a tabular output with the following columns:

Column Name	Description
readid	Unique identifier for the read.
signalid	Identifier for the signal corresponding to the read.
start	Start position of the signal segment in the read.
end	End position of the signal segment in the read.
basepos	Reference base position in the genomic sequence.
base	The detected base at this position.
motif	The surrounding sequence motif in which the base appears.
state	The methylation state (or modification state) of the base.
posterior_probability	Probability assigned to the predicted segment.
polish	Polished kmer, only available in resquiggle mode.

Example Output

Below is an example of the output generated by Dynamont:

readid,signalid,start,end,basepos,base,motif,state,posterior_probability,polish
476b4ed2-7865-4f81-9f78-82d614fb40a2,476b4ed2-7865-4f81-9f78-82d614fb40a2,12762,12777,53,A,AAAAAAAAA,M,0.12434,NA
476b4ed2-7865-4f81-9f78-82d614fb40a2,476b4ed2-7865-4f81-9f78-82d614fb40a2,12777,12791,52,A,AAAAAAAAA,M,0.12146,NA
476b4ed2-7865-4f81-9f78-82d614fb40a2,476b4ed2-7865-4f81-9f78-82d614fb40a2,12791,12806,51,A,AAAAAAAAA,M,0.11881,NA
476b4ed2-7865-4f81-9f78-82d614fb40a2,476b4ed2-7865-4f81-9f78-82d614fb40a2,12806,12820,50,A,AAAAAAAAA,M,0.11665,NA

Exit-Codes

• -11: Segmentation fault
• -9: Out of Memory error. Decrease the number of processes or move to a system with more memory.
• -6: std::bad_alloc
• 1: resquiggle mode specific: alignment score (Z) does not match between forward and backward run in preprocessing on signal (T) and read (N).
• 2: resquiggle mode specific: alignment score (Z) does not match between forward and backward run in preprocessing on signal (T) and error correction (C).
• 3: Alignment score (Z) does not match between forward and backward pass or is -Infinity
• 4: Input signal is missing or not found in stdin stream
• 5: Input read is missing or not found in stdin stream
• 6: raw file does not exist
• 7: Invalid model path was provided
• 8: Provided ONT signal is too short
• 9: Read is too short
• 10: Signal is smaller than read
• 11: Read is smaller than kmerSize of provided pore model
• 20: Terminated using KeyboardInterrupt (Ctrl + C)