ALDEx2-python

Run a popular ALDEx2 tool in Python via rpy2.

This tool is aware of the compositionality of sequencing data, and treats them accordingly.

Dependencies

Create a new conda environment and install required dependencies:

conda create --name pyaldex2 -c conda-forge -c bioconda bioconductor-aldex2 rpy2 pandas

Activate the environment:

conda activate pyaldex2

Run it

Clone this repository to a convinient location to store the script:

git clone https://github.com/trickovicmatija/ALDEx2-python.git

Import the module and set the path:

import pandas as pd
import sys, os

sys.path.append("<CHANGE_PATH>/ALDEx2-python") # Append the location of cloned repository
import pyaldex2

path = f'{os.path.dirname(pyaldex2.__file__)}/run_aldex2.R' # Set the path of the "run_aldex2.R" R-script. Default: same directory as the Python module.

Import test data:

counts = pd.read_csv("<CHANGE_PATH>/ALDEx2-python/test_data/raw_counts.tsv",sep='\t',index_col=0) # It will automatically orient the dataframe
metadata = pd.read_csv("<CHANGE_PATH>/ALDEx2-python/test_data/metadata.tsv",sep='\t',index_col=0)

Run the script:

result_df = pyaldex2.run_aldex2(counts,metadata,'t',r_script_path=path)

For the description of ALDEx2 output see here.

You can also get CLR-transformed data, which represent a median between all created Monte-Carlo instances:

clr_data = pyaldex2.get_clr(counts,metadata,mc_samples=16,r_script_path=path)

Or, you can get only one instance:

clr_instance = pyaldex2.get_clr_instance(counts,metadata,instance=1,mc_samples=16,r_script_path=path)

Plot

Before plotting, make sure neccessary libraries (numpy, seaborn, matplotlib.pyplot) are installed.

If not, install them in order to continue.

You can create a MA-plot using pyaldex2.MA_plot():

pyaldex2.MA_plot(result_df,effect_threshold=1)

Check help of the function for details.

Also, you can create a vulcano-plot using pyaldex2.vulcano_plot():

pyaldex2.vulcano_plot(result_df)

Check help of the function for details.

Disclaimer

This script is only running the original ALDEx2 tool in Python. I don't take any responsibility or credits from it. For any questions about the tool itself, please write to them.