Skip to content

Latest commit

 

History

History
65 lines (44 loc) · 2.27 KB

README.md

File metadata and controls

65 lines (44 loc) · 2.27 KB
Raw

SMOTE_DNA

SMOTE_DNA is a Synthetic Minority Over-sampling TEchnique (SMOTE)-based adapted to generate synthetic nucleotide sequence data.

By following these instructions, you should be able to generate synthetic nucleotide sequences using SMOTE_DNA. For more details on the different modes and algorithm-specific configurations, please refer to the original research publication.

Table of Contents

Setting Up the Environment

First, create the smote_dna CONDA environment:

conda create python=3.10.11 --name smote_dna -y

Activate the smote_dna environment:

conda activate smote_dna

Then, install the required dependencies:

pip install imblearn

Preparing the Input Data

The input should be a CSV file with two columns:

  • The first column contains the nucleotide sequences.
  • The second column contains the class labels for each sequence.

You can refer to the example file located at input/example_smote_input.csv for the expected format.

Here is an example command to create a CSV file in the required format from an existing dataset. This command replaces all instances of "U" with "T" in the sequences. The shuffle command is optional.

awk 'BEGIN { FS = ","; OFS="," } { gsub("U","T",$2); print $2, $3 }' input/rna_sequences.csv | shuf > input/example_smote_input.csv

Generating Synthetic Data

To generate synthetic data using SMOTE_DNA, execute the following command. Don't forget to activate the smote_dna environment.

python smote_dna/smote_dna.py --mode A --samples 1000 --input input/example_smote_input.csv --output output/example_smote_output.txt

Where:

  • mode: choose between modes A, B, or C. Refer to the corresponding research paper to understand the differences between the modes;
  • samples: the number of synthetic samples to generate;
  • input: the path to the CSV file containing the real nucleotide sequence data (formatted as described above);
  • output: the path where the generated synthetic data will be saved.

The command above will generate 1000 synthetic samples and save them in the output/example_smote_output.txt file.