Measuring the reproducibility and quality of Hi-C data

Abstract

BackgroundHi-C is currently the most widely used assay to investigate the 3D organization of the genome and to study its role in gene regulation, DNA replication, and disease. However, Hi-C experiments are costly to perform and involve multiple complex experimental steps; thus, accurate methods for measuring the quality and reproducibility of Hi-C data are essential to determine whether the output should be used further in a study.ResultsUsing real and simulated data, we profile the performance of several recently proposed methods for assessing reproducibility of population Hi-C data, including HiCRep, GenomeDISCO, HiC-Spector, and QuASAR-Rep. By explicitly controlling noise and sparsity through simulations, we demonstrate the deficiencies of performing simple correlation analysis on pairs of matrices, and we show that methods developed specifically for Hi-C data produce better measures of reproducibility. We also show how to use established measures, such as the ratio of intra- to interchromosomal interactions, and novel ones, such as QuASAR-QC, to identify low-quality experiments.ConclusionsIn this work, we assess reproducibility and quality measures by varying sequencing depth, resolution and noise levels in Hi-C data from 13 cell lines, with two biological replicates each, as well as 176 simulated matrices. Through this extensive validation and benchmarking of Hi-C data, we describe best practices for reproducibility and quality assessment of Hi-C experiments. We make all software publicly available at http://github.com/kundajelab/3DChromatin_ReplicateQC to facilitate adoption in the community.Electronic supplementary materialThe online version of this article (10.1186/s13059-019-1658-7) contains supplementary material, which is available to authorized users.

FreeHi-C: high fidelity Hi-C data simulation for benchmarking and data augmentation

“…FreeHi-C takes as input raw Hi-C sequencing reads in FASTQ format and estimates the frequency of genomic fragment interactions. This is fundamentally different from existing methods that simulate Hi-C contact matrices under a series of assumptions 9,11,14,15 . Subsequently, FreeHi-C generates pairs of sequencing reads that represent the interacting fragment pairs with embedded random nucleotide mutations and indels while conserving the proportion of chimeric reads.…”

“…Recent maturation of chromosome conformation capture (3C) 1 and Hi-C sequencing technologies 2,3 led to high-throughput profiling of three-dimensional chromatin architecture and revealed transformative insights on long-range gene regulation [4][5][6] . Alongside the technological breakthroughs, a growing number of methodologies and algorithms [7][8][9][10][11][12][13][14][15][16] emerged for the analysis of Hi-C and other 3C-derived data types. These methods are developed and benchmarked on disparate biological and simulated or computationally-constructed datasets that are often customized for the methods under the study.…”

“…Benchmarking and evaluation of Hi-C methods have relied on either conducting distancestratified permutations of the contact matrices 15 to nullify the underlying biological structure or simulating contact matrices directly by capturing only the most general spatial structures for specific analysis purposes 9,11,13,14,18,19 . Additionally, pooling biological replicates and then randomly partitioning to generate pseudo-replicates and downsampling are also common approaches to study similarity metrics or evaluate the sequencing depth effect 12,15,18,19 . A more systematic Hi-C simulation method, Sim3C 20 , has been recently proposed for the design of Hi-C experiments with respect to the selection of restriction enzyme, determination of sequencing depth, and power analysis.…”

Activity-by-Contact model of enhancer specificity from thousands of CRISPR perturbations

“…To do so, we first identified alternative ways to estimate Contact in the ABC model; although maps of chromatin accessibility and histone modifications are available in many cell types, maps of 3D contacts are not. Because contact frequencies in Hi-C data correlate well across cell types (Note S3) (38,39), we compared versions of the ABC model in which we estimated Contact for each DE-G pair using either K562 Hi-C data or the average Hi-C contact frequency from 8 other human cell types. Both approaches performed similarly at predicting our CRISPR data in K562 cells (AUPRC = 0.66 and 0.68 respectively; Fig.…”

scHiCTools: a computational toolbox for analyzing single-cell Hi-C data

“…Previously, similarity measures for comparing Hi-C contact matrices mostly focus on bulk Hi-C data (Yardımcı et al, 2019). These methods evaluate how likely two bulk Hi-C experiments are generated from the same biological sample.…”