VEHiCLE Variationally Encoded Hi-C Loss Enhancement

This directory contains code used in running experiments for the paper:

Highsmith, M. & Cheng, J. VEHiCLE: a Variationally Encoded Hi-C Loss Enhancement algorithm. doi:10.1101/2020.12.07.413559.

1. Contact:

Max Highsmith
Department of Computer Science
Email mrh8x5@mail.missouri.edu

2. Content of Folders:

Data:    The Raw Data, Data Loaders, and preprcoessing scripts
Models:  Pytorch implementation of models used in experiments
Weights: Trained weights of experiments
Experiments: Scripts used to run experiments
Fig_Scripts: Scripts used To generate Figures
other_tools: tools built by other labs need to run experiments

3. Hi-C Data used in this study:

In our study we used Hi-C data from GSE63525.  Datasets are programatically downloaded and formatted via the dataloader objects (Data/<cell_line>_DataModule.py) but can be found in their raw format at
*https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE63525*.

3. Usage

To view the interactive tunable Hi-C Contact matrix generating GUI run

python Generative_GUI.py

To enhance your own HiC data run

Enhance_Your_Own_Data.py

You will need to edit the configuration variables:

1. YOUR_CELL_LINE : "name of Cell line"
2. LOW_RES_HIC : "location of hic data"
3. CHRO : "chromosome number to be inspected"

This will extract Low Res HiC Contact Matrices in "<Your_Line>/Full_Mats" and will place an enhanced matrix in "<Your_Line>/Full_Enhanced"

To obtain the Insulation Score Identified TAD boundaries run

python Insulation.py enhanced_cell_line chromosome coordinate_file resolution tadfn

example

python Insulation.py GM12878/Full_Enhanced/full_enh.npy 7 GM12878/Full_Mats_Coords/coords_chr7_res_10000.npy 10000 enh.txt

4. Reproducibility

Because of their size we store our models on our lab server. The trained models used in experiments can be found at http://sysbio.rnet.missouri.edu/3dgenome/VEHiCLE_Weights/