New R Package ChIPSP Maps Gene Regulation in 3D, Revealing Hidden Cancer Targets

For more than a decade, chromatin immunoprecipitation followed by sequencing, better known as ChIP-seq, has been the workhorse method for locating where transcription factors bind across the genome. The technique generates maps of protein-DNA interactions at extraordinary resolution, allowing researchers to identify the precise landing sites of regulatory proteins on chromatin. Yet a stubborn problem has persisted in how these binding maps are translated into lists of candidate target genes. Conventional downstream annotation typically assigns a binding peak to the nearest gene along the linear sequence of the genome, an approach that implicitly assumes regulatory regions act on their closest linear neighbors. In reality, the genome is not a flat string of letters but a folded, looped, three-dimensional structure, and enhancers frequently contact the promoters of genes located tens or even hundreds of kilobases away by looping through nuclear space. A new open-source tool aims to close this gap between one-dimensional annotation and three-dimensional biology.

The tool, called Spatial ChIP, or ChIPSP, is an R package developed by Tianyi Zhou, Kevin Song, Hui Huang, and Qin Feng of the Center for Nuclear Receptors and Cell Signaling and the Department of Biology and Biochemistry at the University of Houston, together with Ning Lyu of the Division of Pharmacoepidemiology and Pharmacoeconomics and the Harvard-MIT Center for Regulatory Science at Harvard Medical School. Described in the journal BMC Genomics, ChIPSP integrates standard ChIP-seq data with chromatin loop interactions derived from Hi-C, a genome-wide chromosome conformation capture assay that detects which DNA segments physically come into contact inside the nucleus. By overlaying transcription factor binding peaks onto Hi-C maps, the package prioritizes candidate transcription factor-associated genes that lie within the same three-dimensional regulatory neighborhoods, even when they are far apart in linear genomic distance.

The technical logic behind ChIPSP is straightforward but powerful. Hi-C experiments, often performed at resolutions ranging from tens of kilobases down to a few kilobases, yield millions of pairwise chromatin interactions that can be filtered into significant loop calls. ChIPSP takes these loop anchors and asks which of them overlap transcription factor binding sites identified by ChIP-seq, and which of the opposing anchors contain gene promoters. A gene becomes a candidate target when its promoter participates in a loop whose other anchor is occupied by the transcription factor of interest. This spatial framework captures enhancer-promoter communication that crosses chromatin loop boundaries, situations in which the nearest-gene heuristic would either pick the wrong gene or miss the true regulatory relationship entirely. The result is a ranked table of candidate genes accompanied by the supporting loop and peak evidence, allowing researchers to inspect the underlying interactions for any locus of interest.

To evaluate the package, the team applied it to one of the most clinically consequential transcription factors in oncology: the androgen receptor, or AR, the nuclear hormone receptor that drives prostate cancer growth and is the principal target of modern androgen receptor pathway inhibitors such as darolutamide. Using ChIP-seq data from LNCaP prostate cancer cells, a widely used model of androgen-dependent prostate cancer, ChIPSP identified 1,499 candidate AR-associated genes. Strikingly, 658 of these genes, more than forty percent of the candidate list, were missed entirely by conventional linear annotation methods. In other words, more than four hundred additional potential AR targets emerged simply by asking how the genome folds rather than how it reads in a straight line.

The biological plausibility of these newly detected candidates was tested using RNA-seq. Many of the 658 spatially linked genes proved to be androgen-responsive, changing their expression in response to androgen signaling, which is exactly what would be expected of genuine AR targets. Pathway analysis of the expanded candidate set revealed enrichment in developmental transcriptional programs that were distinct from the pathways captured by standard ChIP-seq annotation alone. This suggests that conventional approaches have not merely been undercounting targets but may have been systematically skewing the perceived biology of androgen receptor signaling toward a subset of programs that happen to sit near AR binding sites in linear space.

Among the individual loci highlighted by the analysis, the genes KRT8, which encodes the keratin 8 intermediate filament protein, and MAF, which encodes a transcription factor implicated in cancer biology, stood out as consistent with candidate long-range AR-associated regulation across chromatin loop boundaries. At these loci, AR-bound distal regions are brought into contact with the gene promoters by looping, providing a structural explanation for how the receptor could control genes that appear distant or unrelated in linear genomic coordinates. Supplementary analyses further supported the spatial linkages, showing androgen receptor occupancy and H3K27ac enhancer marks, a histone modification associated with active regulatory elements, at the relevant genomic regions, and connecting distal AR-bound sites to additional targets including NDRG1 and ERRFI1.

To demonstrate that the approach is not specific to a single receptor or cell type, the researchers applied ChIPSP to the glucocorticoid receptor, or GR, another nuclear hormone receptor, using ChIP-seq data from A549 lung cancer cells. In this second system, the package nominated candidate gene targets including IRS2, UBL3, and FOXO1. The team then showed, using RT-qPCR validation experiments, that these genes were responsive to dexamethasone, a synthetic glucocorticoid that activates the receptor. The confirmation that spatially nominated candidates respond experimentally to receptor activation provides a second, independent nuclear-receptor example and strengthens the case that the method captures genuine regulatory relationships rather than genomic coincidences.

The implications for cancer research are considerable. Nuclear receptors such as AR and GR are among the most heavily drugged classes of transcription factors, and identifying their full repertoire of target genes is essential for understanding both therapeutic response and resistance. Genes that regulate cell identity, survival, and lineage plasticity often sit far from the binding sites that control them, and a nearest-gene annotation strategy can leave entire regulatory programs unexamined. By bridging protein-DNA binding data with chromatin interaction maps, ChIPSP extends ChIP-seq annotation into a spatial framework and complements rather than replaces conventional peak-to-gene annotation. Because it is distributed as an R package, the tool is accessible to the same bioinformatics community that already processes ChIP-seq data, requiring only ChIP-seq peak calls and publicly available Hi-C loop datasets as input.

The work also arrives at a moment when three-dimensional genome organization has moved from a specialist interest to a mainstream concern in genomics. Hi-C and related chromatin conformation capture technologies have matured, and public repositories now hold chromatin loop maps for many of the same cancer cell lines routinely used in ChIP-seq studies. Tools like ChIPSP make it practical to combine these data layers routinely rather than treating 3D context as an afterthought reserved for detailed single-locus case studies. The authors note that Hi-C loop detection is resolution-dependent, an important technical caveat for users, since loops called at coarse resolution may merge distinct regulatory contacts, while very high resolution data may be needed to resolve fine-scale enhancer-promoter architecture.

For researchers studying transcription factors in cancer and beyond, ChIPSP offers a practical route to regulatory targets that are invisible to conventional approaches. The package’s developers, who received no external funding for the research, have made the tool available as an open-access contribution, with the underlying study published in BMC Genomics and accompanied by ranked candidate gene lists, pathway analyses across KEGG, Gene Ontology, Hallmark, and Reactome databases, and experimental validation data. As spatially aware annotation becomes part of the standard ChIP-seq analysis pipeline, the field may find that the genomes of cancer cells have been holding far more regulatory information in their folds than linear analysis ever revealed.

Subject of Research: An R package integrating ChIP-seq with Hi-C chromatin loop data to characterize spatial transcription factor gene regulation in three-dimensional genomic context.

Article Title: Spatial ChIP (ChIPSP), an R package for characterizing spatial gene regulation

Article References: Spatial ChIP (ChIPSP), an R package for characterizing spatial gene regulation. (n.d.). https://doi.org/10.1186/s12864-026-13371-w

Image Credits: AI Generated

DOI: 10.1186/s12864-026-13371-w

Keywords: ChIPSP, ChIP-seq, Hi-C, three-dimensional genome organization, chromatin looping, transcription factor, androgen receptor, glucocorticoid receptor, prostate cancer, enhancer-promoter interactions, R package, gene regulation