Abstract
Hi-C experiments have been widely adopted to study chromatin spatial organization, which plays an essential role in genome function. We have recently identified frequently interacting regions (FIREs) and found that they are closely associated with cell-type-specific gene regulation. However, computational tools for detecting FIREs from Hi-C data are still lacking. In this work, we present FIREcaller, a stand-alone, user-friendly R package for detecting FIREs from Hi-C data. FIREcaller takes raw Hi-C contact matrices as input, performs within-sample and cross-sample normalization, and outputs continuous FIRE scores, dichotomous FIREs, and super-FIREs. Applying FIREcaller to Hi-C data from various human tissues, we demonstrate that FIREs and super-FIREs identified, in a tissue-specific manner, are closely related to gene regulation, are enriched for enhancer-promoter (E-P) interactions, tend to overlap with regions exhibiting epigenomic signatures of cis-regulatory roles, and aid the interpretation of GWAS variants.
Introduction: The Need for Tissue-Specific Genomic Features
The study of the 3D genome has traditionally focused on structural units like A/B compartments (large-scale active/inactive zones) and Topologically Associating Domains (TADs). While these features are fundamental, they are often "evolutionarily conserved" or consistent across different cell types. This poses a challenge for researchers trying to understand why a specific gene is active in one tissue but silenced in another.
Frequently Interacting Regions (FIREs) represent a more dynamic layer of genome organization. These genomic neighborhoods (typically 40Kb) show significantly higher local interaction frequencies than their neighbors and are often enriched with tissue-specific enhancers. FIREcaller bridges the gap between raw data and biological insight by providing a standardized, stand-alone R package for identifying these hotspots.
The FIREcaller Pipeline: From Raw Data to Regulatory Hubs
FIREcaller transforms raw Hi-C contact matrices into statistically significant interaction maps through a multi-step pipeline:
- Local Interaction Mapping: The tool calculates the total number of local interactions (within <200Kb) for each genomic locus (defaulting to 40Kb bins).
- Within-Sample Normalization: Using the
HiCNormCismethod, FIREcaller employs Poisson regression to filter out systematic biases such as GC content, mappability, and restriction enzyme cutting sites. - Cross-Sample Normalization: Quantile normalization ensures scores are comparable across different tissues or replicates.
- Statistical Significance: FIRE scores are converted to Z-scores and p-values (p < 0.05 for FIRE identification).
- Super-FIRE Detection: Adjacent FIREs are merged to identify "super-FIREs"—massive hubs of local chromatin activity, analogous to super-enhancers.
Validating Functional Significance
The authors validated FIREcaller using high-profile human Hi-C datasets, including human hippocampus tissue and cortical tissue across developmental epochs (fetal vs. adult).
Key findings include:
- Developmental Dynamics: The identification of ~4,815 differentially regulated FIREs between fetal and adult brain stages highlights a massive "rewiring" of the genome during maturation.
- Gene Expression Correlation: Genes near fetal-specific FIREs were significantly up-regulated in fetal tissue (p < 2.2e-16), proving that FIREcaller identifies regions with actual functional impact.
- Regulatory Hubs: FIREs are significantly enriched for tissue-specific enhancers, acting as major regulatory hubs for genes essential to tissue identity.
Clinical Relevance and Case Study
FIREcaller is particularly useful for interpreting the "dark matter" of the genome—non-coding regions where most disease-associated genetic variants (SNPs) reside. By determining if these variants sit inside a FIRE, researchers can hypothesize how a mutation might disrupt a 3D interaction hub.
In the hippocampus, FIREcaller identified a 400Kb super-FIRE overlapping the TCF4 gene, a known player in neurodevelopment, and a schizophrenia-associated SNP (rs9960767). This provides a concrete structural mechanism for how this genetic risk factor might function.
Conclusion and Future Directions
While FIREcaller provides a robust framework, its performance is tied to the resolution of the input Hi-C data. As ultra-high-resolution technologies like Micro-C become more common, future iterations may allow for even finer-scale detection of interaction hotspots. Currently, FIREcaller stands as a vital tool for the bioinformatics community to explore the tissue-specific 3D genome.
Summary updated