/SSG-LUGIA

Primary LanguagePython

SSG-LUGIA

Single Sequence based Genome Level Unsupervised Genomic Island Prediction Algorithm

This repository contains the original implementation of SSG-LUGIA, an unsupervised learning based tool to predict genomic islands.

Project Website

SSG-LUGIA Project Website

Publication

Codes

The codes for SSG-LUGIA are written in python and can be found here

Requirements

Note : In the latest version of Scikit-Learn the implementation of EllipticEnvelope has been changed, so please use the specified version to obtain reproducible results.

  • numpy==1.17.0
  • biopython==1.70
  • tqdm==4.19.5
  • scikit-learn==0.19.1

Usage

  1. Clone this repository

$ git clone https://github.com/nibtehaz/SSG-LUGIA.git

  1. Install the requirements

$ pip3 install -r requirements.txt

  1. Navigate to the /codes directory

  2. Launch Python CLI

$ python3

  1. Import the SSG-LUGIA pipeline

from main import SSG_LUGIA

  1. Execute it with a genome sequence fasta file and a standard model name from SSG-LUGIA-F, SSG-LUGIA-R, SSG-LUGIA-P

SSG_LUGIA(sequence_fasta_file_path='sample_data/NC_003198.1.fasta',model_name='SSG-LUGIA-F')

  1. Alternatively, the model name can be omitted and the user can set the parameters interactively

SSG_LUGIA(sequence_fasta_file_path='sample_data/NC_003198.1.fasta')

  1. Alternatively, the user can input a custom model as dictionary

SSG_LUGIA(sequence_fasta_file_path='sample_data/NC_003198.1.fasta',model_parameters=custom_model)

  1. Alternatively, the user can create a model based on their requirement, save it as a json file and input the path to the json file

SSG_LUGIA(sequence_fasta_file_path='sample_data/NC_003198.1.fasta',model_name='path-to-json')

Model Parameters

SSG-LUGIA combines several sequence based features to infer GIs using an unsupervised anomaly detection pipeline. The various model parameters can be found in SSG-LUGIA Model Parameters. Users can develop custom model variants by changing these parameters and also save the model as json for future use.

Citation Request

If you use SSG-LUGIA in your project, please cite the following paper

@article{ibtehaz2021ssg,
  title={SSG-LUGIA: Single Sequence based Genome Level Unsupervised Genomic Island Prediction Algorithm},
  author={Ibtehaz, Nabil and Ahmed, Ishtiaque and Ahmed, Md Sabbir and Rahman, M Sohel and Azad, Rajeev K and Bayzid, Md Shamsuzzoha},
  journal={Briefings in Bioinformatics},
  year={2021}
}