Detecting chromosomal interactions in Capture Hi-C data with CHiCAGO and companion tools

Abstract

Competing interests P.F.P. is currently an employee of Inivata Limited. J.C. is currently an employee of AstraZeneca and may or may not own stock options. M.S. is a cofounder of Enhanc3D Genomics Ltd. The rest of the authors declare no competing interests.

High-resolution promoter interaction analysis in Type 3 Innate Lymphoid Cells implicates Batten Disease geneCLN3in Crohn’s Disease aetiology

“…1B ; Tables S1-S2 ). Binning resulted in the detection of longer-range contacts, as we reported previously in other cell types 37 ( Fig. 1C, D ).…”

“…1A ). Significant promoter contacts were detected with CHiCAGO 36 at a single-fragment resolution, as well as after pooling the ‘other end’ fragments into ∼5kb bins, while leaving the baited promoter-containing fragment unbinned 37 . Using this approach, we detected 31,003 contacts between promoters and promoter-interacting regions (PIRs) at a single-fragment resolution and 58,632 contacts in 5kb bins ( Fig.…”

“…Here, we address this limitation by using a high-resolution and efficient PCHi-C protocol to profile the cis -regulatory wiring of ILC3s isolated from human tonsils 29 . We detect promoter-enhancer contacts in PCHi-C data using a combination of our established statistical interaction-calling methodology (CHiCAGO) 36,37 and a PCHi-C-adapted Activity-by-Contact (ABC) 13,38 approach devised in this study. We develop a modified PCHi-C-aware GWAS gene prioritisation algorithm that incorporates multivariate statistical fine-mapping (multiCOGS) and use it to prioritise known and novel genes for CD through chromatin contacts in ILC3s, including the CLN3 gene, mutations in which underpin ∼80% cases of the neurodegenerative disorder Batten disease 39,40 .…”

Prioritisation of Candidate Genes Underpinning COVID-19 Host Genetic Traits Based on High-Resolution 3D Chromosomal Topology

“…The cell-type specificity of COGS scores may not be consistent with the expression patterns of the prioritised genes, while using COGS in a pooled setting across multiple samples increases the sensitivity of the analysis ( Javierre et al, 2016 ). At the same time, the coverage and design of different PCHi-C datasets may have systematic effects on detected interaction signals ( Freire-Pritchett et al, 2021 ). Therefore, COGS was run separately for data from each GWAS meta-analysis using a pool of promoter interactions with CHiCAGO ​​scores ( Cairns et al, 2016 ) above 5 in at least one cell type in either dataset (Javierre: 707,583 interactions involving 21,102 baited promoter fragments; Ho: 43,265 interactions involving 9,955 baited promoter fragments).…”

Genetic variation in correlated regulatory region of Immunity

“…We found that the fraction of correlated chromatin peak pairs (per chromosome pair-wise association testing, FDR 1%) increased with the PCHi-C interaction score (PCHi-C interactions are considered significant if the CHiCAGO score≥ 5) [29] [Supplementary Figure 3A]. Moreover, correlated histone peak pairs were more likely to be in close 3D physical proximity (CHiCAGO score ≥ 5) than uncorrelated peaks, with the effect maximizing for pairs separated by 50-500kb [Figure 3A and Supplementary 3B,C] (for distances ¡20kb, we expect genomic distance noise to affect short-range PCHi-C interactions [30], therefore we do not consider this distance interval in the analysis). Then, we explored how cis gene-CRD associations may reflect long-range physical interactions between regulatory elements and gene promoters.…”