liuchuwei / pum6a

pum6a

Dependence

Table of Contents

1. Intallation
2. Usage
3. References
4. License
5. Contact

Intallation

1.Clone the project

git clone https://github.com/liuchuwei/pum6a.git

2.Install conda environment

conda create -n pum6a python=3.8
conda activate pum6a
pip install torch==2.2.1 torchvision==0.17.1 torchaudio==2.2.1 --index-url https://download.pytorch.org/whl/cu118
pip install dask==2023.5.0 h5py==3.10.0 numpy==1.24.3 pandas==2.0.3 scikit-learn==1.3.2 tqdm==4.66.1 toml==0.10.2 statsmodels==0.14.1

The usage of pum6a for m6A detection require the tombo environment.

You can install MINES, m6Anet, ELIGOS, Nanom6A, and Epinano environment according to your need. Usage example can be found in the m6a_detection directory.

3.prepare tookit: check and modify the tool paths of tookit.py file (in 'utils' directory).

Usage

Experiment

Experiment of pum6a framework for different positive and unlable bags datasets.

python run.py experiment --config $*.toml

for example: python run.py experiment --config log/Internet_pum6a_0.5Freq_88888.toml

m6A modification detection

1.Basecalling

python process/01.basecalling.py -i $fast5 -o $out

2.Resguiggle

preprocess

conda activate tombo
python process/02.resquiggle_pre.py -f $fast5 -o $out

annotate_raw_with_fastqs

cat *.fastq > merge.fastq
python process/03.resquiggle.py preprocess annotate_raw_with_fastqs \
--fast5-basedir $single \
--fastq-filenames $merge_fastq \
--overwrite \
--processes 8

resquiggling

 python process/3.resquiggle.py resquiggle $fast5 $reference \
 --rna \
 --corrected-group RawGenomeCorrected_000 \
 --basecall-group Basecall_1D_000 \
 --overwrite \
 --processes 16 \
 --fit-global-scale \
 --include-event-stdev

3.Minimap

python process/04.minimap.py -i <directory of fastq files> -o <output directory> -r <path of reference>

4.m6a detection

4.1 activate environment

conda activate pum6a

4.2 preprocess

python run.py preprocess --single $single_fast5 -o $output -g $genome.fa -r $transcript.fa -i $gene2transcripts.txt -b $bam

4.3 train model

python run.py train --config $config.toml

4.4 predict

python run.py predict --config $config.toml

4.5 evaluate

python run.py evaluate --config $config.toml

License

Distributed under the MIT License. See LICENSE for more information.

Contact

[email protected]

Reference