TAD-Lactuca

TAD-Lactuca is a tool to predict Topologically Associating Domains(TAD) boundary using histone marks information. It was written in Python language, using Random Forests(RF) and Multilayer Perception(MLP).

Overview

The data fold contains the origin data (expect the bigwig file, the detail can be found in paper)
The src fold contains all scripts, such as data processing, modeling and plot.
The cache fold contains the main result and some processed the multiple resolution feature signal, such as the bin's length range(40kb,20kb,10kb), the bin's number range(10,8,6).
The pic fold contains the all signal pattern of different feature.

Dependency

Python3.*
Numpy
Pandas
Tensorflow / Tensorflow-gpu
Keras
Scikit-Learn
Matplotlib

Preprocessing

The data fold contains the original files, the signals of different type histone were calculate by bedtool, the details can be seen from ./src/utils/get_signal_plot.py

Usage

If you want to use the example data, you can run the script as:

python3 tad_lactuca.py

If you want to use your own data, please run the script as:

python3 tad_lactuca.py [the path of your data]

Input

The feature data of the locus you want to know is boundary or not
Only the base location you want to know is in the domain or boundary

Output

The probability of the input data belong to each category.
The signal of the locus.