HiCNN2: Enhancing the Resolution of Hi-C Data Using an Ensemble of Convolutional Neural Networks.

Tong Liu,Zheng Wang

doi:10.3390/genes10110862

Abstract

We present a deep-learning package named HiCNN2 to learn the mapping between low-resolution and high-resolution Hi-C (a technique for capturing genome-wide chromatin interactions) data, which can enhance the resolution of Hi-C interaction matrices. The HiCNN2 package includes three methods each with a different deep learning architecture: HiCNN2-1 is based on one single convolutional neural network (ConvNet); HiCNN2-2 consists of an ensemble of two different ConvNets; and HiCNN2-3 is an ensemble of three different ConvNets. Our evaluation results indicate that HiCNN2-enhanced high-resolution Hi-C data achieve smaller mean squared error and higher Pearson’s correlation coefficients with experimental high-resolution Hi-C data compared with existing methods HiCPlus and HiCNN. Moreover, all of the three HiCNN2 methods can recover more significant interactions detected by Fit-Hi-C compared to HiCPlus and HiCNN. Based on our evaluation results, we would recommend using HiCNN2-1 and HiCNN2-3 if recovering more significant interactions from Hi-C data is of interest, and HiCNN2-2 and HiCNN if the goal is to achieve higher reproducibility scores between the enhanced Hi-C matrix and the real high-resolution Hi-C matrix.

Highlights

The population-cell Hi-C technique [1] can capture genome-wide intra- and inter-chromosomal contacts, which provide proximity information of the DNA and can be used to reconstruct the three-dimensional (3D) structures of chromosomes [2,3,4], define topologically associated domains (TADs) [5,6,7], and reveal significant genomic interactions [8,9]
These results indicate that (1) HiCNN2 consistently performs better than HiCNN and HiCPlus; (2) HiCNN2-1, an improved version of HiCNN, apparently achieves smaller mean squared errors and higher Pearson’s correlations than HiCNN; and (3) it is difficult to distinguish which method is better among the three HiCNN2 architectures as their performances are similar in terms of the MSE
HiCNN2 consists of three different architectures

Summary

Introduction

The population-cell Hi-C technique [1] can capture genome-wide intra- and inter-chromosomal contacts, which provide proximity information of the DNA and can be used to reconstruct the three-dimensional (3D) structures of chromosomes [2,3,4], define topologically associated domains (TADs) [5,6,7], and reveal significant genomic interactions [8,9]. To experimentally obtain high-resolution (e.g., 5 kb) Hi-C data, researchers need to generate more than one billion paired-end reads [9], which may incur a high sequencing cost. Computational methods for resolution enhancement of Hi-C data are indispensable. In this research, we only focus on enhancing population-cell Hi-C data

Methods

Results

Conclusion