The preciseTADworkshop introduces methods for transforming the identification of boundaries that demarcate Topologically Associating Domains (TADs)– referred to as TAD-calling–into a supervised machine learning framework. Chromosome conformation capture technologies combined with high-throughput sequencing (Hi-C) have revealed that chromatin undergoes layers of compaction through DNA looping and folding, forming dynamic 3-dimensional (3D) structures. Among these are TADs, which are known to play critical roles in cell dynamics like gene regulation and cell differentiation. However, precise TAD-calling remains difficult, as it is strongly reliant on Hi-C data resolution. Obtaining genome-wide chromatin interactions at high-resolution is costly resulting in low resolution of Hi-C matrices and high uncertainty in the location of domain boundaries. In this workshop we will circumvent this resolution restriction by building predictive models that leverage high resolution functional genomic element data (ChIP-seq). As an application, we will demonstrate that these methods provide more precise boundary detection compared to a conventional TAD-calling algorithm by evaluating a variety of visualization techniques in relation to the enrichment of key molecular drivers of 3D chromatin. The methods discussed in this workshop will give users tools for bridging the resolution gap between 1D ChIP-seq annotations and 3D Hi-C sequencing data for more precise and biologically meaningful boundary identification.

Key materials for the workshop:

This workshop will be presented at the European Bioconductor Virtual Conference 2020, December 6th, 2020, 5:30pm-6:15pm

# Using Docker

• Pull the latest version of the preciseTADworkshop Docker image, docker pull stilianoudakis/precisetadworkshop:latest
• Run docker run -e PASSWORD=yourpassword -p 8787:8787 -d --rm stilianoudakis/precisetadworkshop:latest. Use -v \$(pwd):/home/rstudio argument to map your local directory to the container.
• Log in to RStudio at http://localhost:8787 using username rstudio and password yourpassword. Note that on Windows you need to provide your localhost IP address like http://191.163.92.108:8787/ - find it using docker-machine ip default in Docker’s terminal.
• Run browseVignettes(package = "preciseTADworkshop"). Click on one of the links, “HTML”, “source”, “R code”.

# Local Installation

if(!require(devtools)) install.packages("devtools")
devtools::install_github(repo = "dozmorovlab/preciseTADworkshop", build_vignettes = TRUE)

If installation fails due to missing packages, install them as follows:

if(!require(BiocManager)) install.packages("BiocManager")
BiocManager::install(c('preciseTAD', 'preciseTADhub'))

# Workshop Description

## Pre-requisites

• Basic knowledge of R syntax and command-line tools
• Familiarity with Hi-C chromatin conformation capture technology
• Understanding of Hi-C data properties (contact matrices, interaction frequencies, Topologically Associating Domains, etc)
• Familiarity with supervised machine learning techniques (e.g. classification) and clustering algorithms

# R / Bioconductor packages used

## Time outline

The workshop duration is ~45 min. The approximate timing of activities are shown below:

Activity Time