Awesome
BOOSTNANO
BOOSTNANO is a tool for preprocessing ONT-Nanopore RNA sequencing reads(before basecalling), it segments/trims the adapter, polyA stall, and the transcription from the raw signal before the basecalling.
BoostNano is part of the Chiron project, if you found BoostNano useful please consider to cite the Chiron paper:
Haotian Teng, Minh Duc Cao, Michael B Hall, Tania Duarte, Sheng Wang, Lachlan J M Coin; Chiron: translating nanopore raw signal directly into nucleotide sequence using deep learning, GigaScience, Volume 7, Issue 5, 1 May 2018, giy037, https://doi.org/10.1093/gigascience/giy037
Installation
git clone https://github.com/haotianteng/BoostNano.git
conda activate YOUR_VIRTUAL_ENVIRONMENT
python BoostNano/setup.py install
Install Pytorch
Code Example
Check out the sample code in the Jupyter Notebook Sample.ipynb for how to use the package.
Usage
Inference
Replace the raw signal in the fast5 files for basecalling (trim the raw signal of adapter and polyA tail):
python boostnano_eval.py -i INPUT_FAST5_FOLDER -o OUTPUT_FOLDER -m MODEL_FOLDER --replace
Or you just want to get the segmentation result in output.csv without modifying the original fast5 files:
python boostnano_eval.py -i INPUT_FAST5_FOLDER -o OUTPUT_FOLDER -m MODEL_FOLDER
A pretrained model is already shipped in boostnan/models, you can use it directly. Notice: This will save the segmented siganl in to the original Signal Slot in the fast5 files, and copy the old signal into the Signal_Old slot, so a basecaller can directly run on the processed fast5 files.
Label&Training
python hand_label.py
Which give you a GUI to label the data.
GUI operation:
S - start
X - skip the current read
Q - quit and save
R - enter into review mode, review the label result or running result by Nanopre
reverse the signal(if already in review mode)
D - go to next signal in review mode
left click - mark a segment
Once you get enough labelled data, use boostnano_train.py to train.
Review
To review the segmentation result, run hand_label.py and enter in to the review mode, and choose the output folder to review the result.
