Skip to content

Latest commit

 

History

History
110 lines (61 loc) · 3.42 KB

README.md

File metadata and controls

110 lines (61 loc) · 3.42 KB

TreasureIsland

TreasureIsland is a machine learning-based Genomic Island prediction software, that uses an unsupervised representation of DNA for its prediction.

TreasureIsland is constructed from the Benbow dataset.

A manuscript is available

Dependencies :

Python >= 3.7 Tested on Linux, mac machine For mac, make sure to run: python3 -m ensurepip --upgrade

Installation:

python3 -m venv venv
source venv/bin/activate

Option1 - Use pip:

python3 -m pip install treasureisland

if treasureisland is already installed :

python3 -m pip install treasureisland --upgrade

Option2 - Locally install package:

git clone https://github.com/priyamayur/GenomicIslandPrediction.git
python3 -m pip install -e GenomicIslandPrediction

Usage:

Option1 - Run TreasureIsland directly from commandline :

Run TreasureIsland from commandline on your DNA fasta file (example DNA files provided here), output is given in csv format:

treasureisland mypath/<DNA file>.fasta [-o <output_file_path>] [-ut <upper threshold value>]

Option2 - Run TreasureIsland from python :

The TreasureIsland package is used to find genomic island predictions which can be downloaded in csv, xlsx, txt file formats demonstrated in TreasureIsland package

Input file:

DNA sequence files in fasta format with a sequenceID.

example: >NC_002620.2 Chlamydia muridarum str. Nigg, complete sequence CACATAGCAAAACACTCAAAGTTTTTCAGCAAAAAAGCTTGTTGAAAAAATTGTTGACCGCCTGTTCACA....

Performance:

TreasureIsland takes 2-5 mins to run depending on the size of the input.

Output :

The results are shown in the following format for each genomic island:

example : NC_002620.2 1.0 130000.0 0.95597

Upper Threshold:

User also has the ability to change the upper threshold value to change the precision and recall tradeoff. upper threshold is set to 0.80 by default.

Example :

treasureisland ecoli.fasta -o gei_output -ut 0.95 Setting the upper threshold to 0.95 would increase the precision and decrease the recall performance.

Testing:

A repository containing some sample DNA files that can be downloaded to test TreasureIsland. Note : github downloads fasta file in txt format (filename.fasta.txt).

example :

treasureisland ecoli.fasta -o gei_output -ut 0.95

Running TreasureIsland package from python:

import the Predictor class from treasureisland package:

from treasureisland.Predictor import Predictor

Instantiate the sequence with the DNA sequence file path as the argument. The DNA file used can be a fasta file.

seq = Predictor("<Path to DNA fasta file>/ecoli.fasta", "<output_file_path>")

Optionally, change the upper threshold value.

seq.change_upper_threshold(0.9)

Get prediction data frame from sequence by running the predict method.

pred = seq.predict()

The predictions can be downloaded in text, csv, excel formats.

seq.predictions_to_csv(pred)
seq.predictions_to_excel(pred)
seq.predictions_to_text(pred)

Contact:

Feel free to contact at banerjee.p1104@gmail.com