Peak Callers Research Articles

Pycallingcards: an integrated environment for visualizing, analyzing, and interpreting Calling Cards data.

Unraveling the transcriptional programs that control how cells divide, differentiate, and respond to their environments requires a precise understanding of transcription factors' (TFs) DNA-binding activities. Calling cards (CC) technology uses transposons to capture transient TF binding events at one instant in time and then read them out at a later time. This methodology can also be used to simultaneously measure TF binding and mRNA expression from single-cell CC and to record and integrate TF binding events across time in any cell type of interest without the need for purification. Despite these advantages, there has been a lack of dedicated bioinformatics tools for the detailed analysis of CC data. We introduce Pycallingcards, a comprehensive Python module specifically designed for the analysis of single-cell and bulk CC data across multiple species. Pycallingcards introduces two innovative peak callers, CCcaller and MACCs, enhancing the accuracy and speed of pinpointing TF binding sites from CC data. Pycallingcards offers a fully integrated environment for data visualization, motif finding, and comparative analysis with RNA-seq and ChIP-seq datasets. To illustrate its practical application, we have reanalyzed previously published mouse cortex and glioblastoma datasets. This analysis revealed novel cell-type-specific binding sites and potential sex-linked TF regulators, furthering our understanding of TF binding and gene expression relationships. Thus, Pycallingcards, with its user-friendly design and seamless interface with the Python data science ecosystem, stands as a critical tool for advancing the analysis of TF functions via CC data. Pycallingcards can be accessed on the GitHub repository: https://github.com/The-Mitra-Lab/pycallingcards.

Analyzing histone ChIP-seq data with a bin-based probability of being signal.

Histone ChIP-seq is one of the primary methods for charting the cellular epigenomic landscape, the components of which play a critical regulatory role in gene expression. Analyzing the activity of regulatory elements across datasets and cell types can be challenging due to shifting peak positions and normalization artifacts resulting from, for example, differing read depths, ChIP efficiencies, and target sizes. Moreover, broad regions of enrichment seen in repressive histone marks often evade detection by commonly used peak callers. Here, we present a simple and versatile method for identifying enriched regions in ChIP-seq data that relies on estimating a gamma distribution fit to non-overlapping 5kB genomic bins to establish a global background. We use this distribution to assign a probability of being signal (PBS) between zero and one to each 5 kB bin. This approach, while lower in resolution than typical peak-calling methods, provides a straightforward way to identify enriched regions and compare enrichments among multiple datasets, by transforming the data to values that are universally normalized and can be readily visualized and integrated with downstream analysis methods. We demonstrate applications of PBS for both broad and narrow histone marks, and provide several illustrations of biological insights which can be gleaned by integrating PBS scores with downstream data types.

Evaluating deep learning for predicting epigenomic profiles.

Deep learning has been successful at predicting epigenomic profiles from DNA sequences. Most approaches frame this task as a binary classification relying on peak callers to define functional activity. Recently, quantitative models have emerged to directly predict the experimental coverage values as a regression. As new models continue to emerge with different architectures and training configurations, a major bottleneck is forming due to the lack of ability to fairly assess the novelty of proposed models and their utility for downstream biological discovery. Here we introduce a unified evaluation framework and use it to compare various binary and quantitative models trained to predict chromatin accessibility data. We highlight various modeling choices that affect generalization performance, including a downstream application of predicting variant effects. In addition, we introduce a robustness metric that can be used to enhance model selection and improve variant effect predictions. Our empirical study largely supports that quantitative modeling of epigenomic profiles leads to better generalizability and interpretability.

Widespread autogenous mRNA-protein interactions detected by CLIP-seq.

Autogenous interactions between mRNAs and the proteins they encode are implicated in cellular feedback-loop regulation, but their extent and mechanistic foundation are unclear. It was recently hypothesized that such interactions may be common, reflecting the role of intrinsic nucleobase–amino acid affinities in shaping the genetic code's structure. Here we analyze a comprehensive set of CLIP-seq experiments involving multiple protocols and report on widespread autogenous interactions across different organisms. Specifically, 230 of 341 (67%) studied RNA-binding proteins (RBPs) interact with their own mRNAs, with a heavy enrichment among high-confidence hits and a preference for coding sequence binding. We account for different confounding variables, including physical (overexpression and proximity during translation), methodological (difference in CLIP protocols, peak callers and cell types) and statistical (treatment of null backgrounds). In particular, we demonstrate a high statistical significance of autogenous interactions by sampling null distributions of fixed-margin interaction matrices. Furthermore, we study the dependence of autogenous binding on the presence of RNA-binding motifs and structured domains in RBPs. Finally, we show that intrinsic nucleobase–amino acid affinities favor co-aligned binding between mRNA coding regions and the proteins they encode. Our results suggest a central role for autogenous interactions in RBP regulation and support the possibility of a fundamental connection between coding and binding.

LanceOtron: a deep learning peak caller for genome sequencing experiments.

MotivationGenome sequencing experiments have revolutionized molecular biology by allowing researchers to identify important DNA-encoded elements genome wide. Regions where these elements are found appear as peaks in the analog signal of an assay’s coverage track, and despite the ease with which humans can visually categorize these patterns, the size of many genomes necessitates algorithmic implementations. Commonly used methods focus on statistical tests to classify peaks, discounting that the background signal does not completely follow any known probability distribution and reducing the information-dense peak shapes to simply maximum height. Deep learning has been shown to be highly accurate for many pattern recognition tasks, on par or even exceeding human capabilities, providing an opportunity to reimagine and improve peak calling.ResultsWe present the peak calling framework LanceOtron, which combines deep learning for recognizing peak shape with multifaceted enrichment calculations for assessing significance. In benchmarking ATAC-seq, ChIP-seq and DNase-seq, LanceOtron outperforms long-standing, gold-standard peak callers through its improved selectivity and near-perfect sensitivity.Availability and implementationA fully featured web application is freely available from LanceOtron.molbiol.ox.ac.uk, command line interface via python is pip installable from PyPI at https://pypi.org/project/lanceotron/, and source code and benchmarking tests are available at https://github.com/LHentges/LanceOtron.Supplementary information Supplementary data are available at Bioinformatics online.

The key role of differential broad H3K4me3 and H3K4ac domains in breast cancer

Epigenetic processes are radically altered in cancer cells. The altered epigenetic events may include histone post-translational modifications (PTMs), DNA modifications, and/or alterations in the levels and modifications of chromatin modifying enzymes and chromatin remodelers. With changes in gene programming are changes in the genomic distribution of histone PTMs. Genes that are poised or transcriptionally active have histone H3 trimethylated lysine 4 (H3K4me3) located at the transcription start site and at the 5′ end of the gene. However, a small population of genes that are involved in cell identity or cancer cell properties have a broad H3K4me3 domain that may stretch for several kilobases through the coding region of the gene. Each cancer cell type appears to mark a select set of cancer-related genes in this manner. In this study, we determined which genes were differentially marked with the broad H3K4me3 domain in normal-like (MCF10A), luminal-type breast cancer (MCF7), and triple-negative breast cancer (MDA-MB-231) cells. We also determined whether histone H3 acetylated lysine 4 (H3K4ac), also a mark of active promoters, had a broad domain configuration. We applied two peak callers (MACS2, PeakRanger) to analyze H3K4me3 and H3K4ac chromatin immunoprecipitation sequencing (ChIP-Seq) data. We identified genes with a broad H3K4me3 and/or H3K4ac domain specific to each cell line and show that the genes have critical roles in the breast cancer subtypes. Furthermore, we show that H3K4ac marks enhancers. The identified genes with the broad H3K4me3/H3K4ac domain have been targeted in clinical and pre-clinical studies including therapeutic treatments of breast cancer.

ChIP-AP: an integrated analysis pipeline for unbiased ChIP-seq analysis.

Chromatin immunoprecipitation coupled with sequencing (ChIP-seq) is a technique used to identify protein–DNA interaction sites through antibody pull-down, sequencing and analysis; with enrichment ‘peak’ calling being the most critical analytical step. Benchmarking studies have consistently shown that peak callers have distinct selectivity and specificity characteristics that are not additive and seldom completely overlap in many scenarios, even after parameter optimization. We therefore developed ChIP-AP, an integrated ChIP-seq analysis pipeline utilizing four independent peak callers, which seamlessly processes raw sequencing files to final result. This approach enables (1) better gauging of peak confidence through detection by multiple algorithms, and (2) more thoroughly surveys the binding landscape by capturing peaks not detected by individual callers. Final analysis results are then integrated into a single output table, enabling users to explore their data by applying selectivity and sensitivity thresholds that best address their biological questions, without needing any additional reprocessing. ChIP-AP therefore presents investigators with a more comprehensive coverage of the binding landscape without requiring additional wet-lab observations.

Peakhood: individual site context extraction for CLIP-seq peak regions.

MotivationCLIP-seq is by far the most widely used method to determine transcriptome-wide binding sites of RNA-binding proteins (RBPs). The binding site locations are identified from CLIP-seq read data by tools termed peak callers. Many RBPs bind to a spliced RNA (i.e. transcript) context, but all currently available peak callers only consider and report the genomic context. To accurately model protein binding behavior, a tool is needed for the individual context assignment to CLIP-seq peak regions.ResultsHere we present Peakhood, the first tool that utilizes CLIP-seq peak regions identified by peak callers, in tandem with CLIP-seq read information and genomic annotations, to determine which context applies, individually for each peak region. For sites assigned to transcript context, it further determines the most likely splice variant, and merges results for any number of datasets to obtain a comprehensive collection of transcript context binding sites. Availability and implementationPeakhood is freely available under MIT license at: https://github.com/BackofenLab/Peakhood.Supplementary information Supplementary data are available at Bioinformatics online.

F-Seq2: improving the feature density based peak caller with dynamic statistics.

Genomic and epigenomic features are captured at a genome-wide level by using high-throughput sequencing (HTS) technologies. Peak calling delineates features identified in HTS experiments, such as open chromatin regions and transcription factor binding sites, by comparing the observed read distributions to a random expectation. Since its introduction, F-Seq has been widely used and shown to be the most sensitive and accurate peak caller for DNase I hypersensitive site (DNase-seq) data. However, the first release (F-Seq1) has two key limitations: lack of support for user-input control datasets, and poor test statistic reporting. These constrain its ability to capture systematic and experimental biases inherent to the background distributions in peak prediction, and to subsequently rank predicted peaks by confidence. To address these limitations, we present F-Seq2, which combines kernel density estimation and a dynamic ‘continuous’ Poisson test to account for local biases and accurately rank candidate peaks. The output of F-Seq2 is suitable for irreproducible discovery rate analysis as test statistics are calculated for individual candidate summits, allowing direct comparison of predictions across replicates. These improvements significantly boost the performance of F-Seq2 for ATAC-seq and ChIP-seq datasets, outperforming competing peak callers used by the ENCODE Consortium in terms of precision and recall.

Comparative analysis of commonly used peak calling programs for ChIP-Seq analysis.

Chromatin immunoprecipitation coupled with high-throughput DNA sequencing (ChIP-Seq) is a powerful technology to profile the location of proteins of interest on a whole-genome scale. To identify the enrichment location of proteins, many programs and algorithms have been proposed. However, none of the commonly used peak calling programs could accurately explain the binding features of target proteins detected by ChIP-Seq. Here, publicly available data on 12 histone modifications, including H3K4ac/me1/me2/me3, H3K9ac/me3, H3K27ac/me3, H3K36me3, H3K56ac, and H3K79me1/me2, generated from a human embryonic stem cell line (H1), were profiled with five peak callers (CisGenome, MACS1, MACS2, PeakSeq, and SISSRs). The performance of the peak calling programs was compared in terms of reproducibility between replicates, examination of enriched regions to variable sequencing depths, the specificity-to-noise signal, and sensitivity of peak prediction. There were no major differences among peak callers when analyzing point source histone modifications. The peak calling results from histone modifications with low fidelity, such as H3K4ac, H3K56ac, and H3K79me1/me2, showed low performance in all parameters, which indicates that their peak positions might not be located accurately. Our comparative results could provide a helpful guide to choose a suitable peak calling program for specific histone modifications.

Improving CLIP-seq data analysis by incorporating transcript information

BackgroundCurrent peak callers for identifying RNA-binding protein (RBP) binding sites from CLIP-seq data take into account genomic read profiles, but they ignore the underlying transcript information, that is information regarding splicing events. So far, there are no studies available that closer observe this issue.ResultsHere we show that current peak callers are susceptible to false peak calling near exon borders. We quantify its extent in publicly available datasets, which turns out to be substantial. By providing a tool called CLIPcontext for automatic transcript and genomic context sequence extraction, we further demonstrate that context choice affects the performances of RBP binding site prediction tools. Moreover, we show that known motifs of exon-binding RBPs are often enriched in transcript context sites, which should enable the recovery of more authentic binding sites. Finally, we discuss possible strategies on how to integrate transcript information into future workflows.ConclusionsOur results demonstrate the importance of incorporating transcript information in CLIP-seq data analysis. Taking advantage of the underlying transcript information should therefore become an integral part of future peak calling and downstream analysis tools.

Galaxy CLIP-Explorer: a web server for CLIP-Seq data analysis.

BackgroundPost-transcriptional regulation via RNA-binding proteins plays a fundamental role in every organism, but the regulatory mechanisms lack important understanding. Nevertheless, they can be elucidated by cross-linking immunoprecipitation in combination with high-throughput sequencing (CLIP-Seq). CLIP-Seq answers questions about the functional role of an RNA-binding protein and its targets by determining binding sites on a nucleotide level and associated sequence and structural binding patterns. In recent years the amount of CLIP-Seq data skyrocketed, urging the need for an automatic data analysis that can deal with different experimental set-ups. However, noncanonical data, new protocols, and a huge variety of tools, especially for peak calling, made it difficult to define a standard.FindingsCLIP-Explorer is a flexible and reproducible data analysis pipeline for iCLIP data that supports for the first time eCLIP, FLASH, and uvCLAP data. Individual steps like peak calling can be changed to adapt to different experimental settings. We validate CLIP-Explorer on eCLIP data, finding similar or nearly identical motifs for various proteins in comparison with other databases. In addition, we detect new sequence motifs for PTBP1 and U2AF2. Finally, we optimize the peak calling with 3 different peak callers on RBFOX2 data, discuss the difficulty of the peak-calling step, and give advice for different experimental set-ups.ConclusionCLIP-Explorer finally fills the demand for a flexible CLIP-Seq data analysis pipeline that is applicable to the up-to-date CLIP protocols. The article further shows the limitations of current peak-calling algorithms and the importance of a robust peak detection.

Population size estimation for quality control of ChIP-Seq datasets.

Chromatin immunoprecipitation followed by sequencing, i.e. ChIP-Seq, is a widely used experimental technology for the identification of functional protein-DNA interactions. Nowadays, such databases as ENCODE, GTRD, ChIP-Atlas and ReMap systematically collect and annotate a large number of ChIP-Seq datasets. Comprehensive control of dataset quality is currently indispensable to select the most reliable data for further analysis. In addition to existing quality control metrics, we have developed two novel metrics that allow to control false positives and false negatives in ChIP-Seq datasets. For this purpose, we have adapted well-known population size estimate for determination of unknown number of genuine transcription factor binding regions. Determination of the proposed metrics was based on overlapping distinct binding sites derived from processing one ChIP-Seq experiment by different peak callers. Moreover, the metrics also can be useful for assessing quality of datasets obtained from processing distinct ChIP-Seq experiments by a given peak caller. We also have shown that these metrics appear to be useful not only for dataset selection but also for comparison of peak callers and identification of site motifs based on ChIP-Seq datasets. The developed algorithm for determination of the false positive control metric and false negative control metric for ChIP-Seq datasets was implemented as a plugin for a BioUML platform: https://ict.biouml.org/bioumlweb/chipseq_analysis.html.

Optimized functional annotation of ChIP-seq data

Different ChIP-seq peak callers often produce different output results from the same input. Since different peak callers are known to produce differentially enriched peaks with a large variance in peak length distribution and total peak count, accurately annotating peak lists with their nearest genes can be an arduous process. Functional genomic annotation of histone modification ChIP-seq data can be a particularly challenging task, as chromatin marks that have inherently broad peaks with a diffuse range of signal enrichment (e.g., H3K9me1, H3K27me3) differ significantly from narrow peaks that exhibit a compact and localized enrichment pattern (e.g., H3K4me3, H3K9ac). In addition, varying degrees of tissue-dependent broadness of an epigenetic mark can make it difficult to accurately and reliably link sequencing data to biological function. Thus, there exists an unmet need to develop a software program that can precisely tailor the computational analysis of a ChIP-seq dataset to the specific peak coordinates of the data and its surrounding genomic features. geneXtendeR optimizes the functional annotation of ChIP-seq peaks by exploring relative differences in annotating ChIP-seq peak sets to variable-length gene bodies. In contrast to prior techniques, geneXtendeR considers peak annotations beyond just the closest gene, allowing users to investigate peak summary statistics for the first-closest gene, second-closest gene, ..., nth-closest gene whilst ranking the output according to biologically relevant events and iteratively comparing the fidelity of peak-to-gene overlap across a user-defined range of upstream and downstream extensions on the original boundaries of each gene's coordinates. We tested geneXtendeR on 547 human transcription factor ChIP-seq ENCODE datasets and 198 human histone modification ChIP-seq ENCODE datasets, providing the analysis results as case studies. The geneXtendeR R/Bioconductor package (including detailed introductory vignettes) is available under the GPL-3 Open Source license and is freely available to download from Bioconductor at: https://bioconductor.org/packages/geneXtendeR/

AIControl: replacing matched control experiments with machine learning improves ChIP-seq peak identification.

ChIP-seq is a technique to determine binding locations of transcription factors, which remains a central challenge in molecular biology. Current practice is to use a ‘control’ dataset to remove background signals from a immunoprecipitation (IP) ‘target’ dataset. We introduce the AIControl framework, which eliminates the need to obtain a control dataset and instead identifies binding peaks by estimating the distributions of background signals from many publicly available control ChIP-seq datasets. We thereby avoid the cost of running control experiments while simultaneously increasing the accuracy of binding location identification. Specifically, AIControl can (i) estimate background signals at fine resolution, (ii) systematically weigh the most appropriate control datasets in a data-driven way, (iii) capture sources of potential biases that may be missed by one control dataset and (iv) remove the need for costly and time-consuming control experiments. We applied AIControl to 410 IP datasets in the ENCODE ChIP-seq database, using 440 control datasets from 107 cell types to impute background signal. Without using matched control datasets, AIControl identified peaks that were more enriched for putative binding sites than those identified by other popular peak callers that used a matched control dataset. We also demonstrated that our framework identifies binding sites that recover documented protein interactions more accurately.

RECAP reveals the true statistical significance of ChIP-seq peak calls.

MotivationChromatin Immunopreciptation (ChIP)-seq is used extensively to identify sites of transcription factor binding or regions of epigenetic modifications to the genome. A key step in ChIP-seq analysis is peak calling, where genomic regions enriched for ChIP versus control reads are identified. Many programs have been designed to solve this task, but nearly all fall into the statistical trap of using the data twice—once to determine candidate enriched regions, and again to assess enrichment by classical statistical hypothesis testing. This double use of the data invalidates the statistical significance assigned to enriched regions, thus the true significance or reliability of peak calls remains unknown.ResultsUsing simulated and real ChIP-seq data, we show that three well-known peak callers, MACS, SICER and diffReps, output biased P-values and false discovery rate estimates that can be many orders of magnitude too optimistic. We propose a wrapper algorithm, RECAP, that uses resampling of ChIP-seq and control data to estimate a monotone transform correcting for biases built into peak calling algorithms. When applied to null hypothesis data, where there is no enrichment between ChIP-seq and control, P-values recalibrated by RECAP are approximately uniformly distributed. On data where there is genuine enrichment, RECAP P-values give a better estimate of the true statistical significance of candidate peaks and better false discovery rate estimates, which correlate better with empirical reproducibility. RECAP is a powerful new tool for assessing the true statistical significance of ChIP-seq peak calls.Availability and implementationThe RECAP software is available through www.perkinslab.ca or on github at https://github.com/theodorejperkins/RECAP.Supplementary information Supplementary data are available at Bioinformatics online.

CD25 (IL2RA) Is a Direct Transcriptional Target of PRDM1 in Activated Human Natural Killer Cells

Background: PRDM1 is a transcription factor that regulates differentiation and/or homeostasis of B and T lymphocytes. It may also play a role in natural killer (NK) cell homeostasis, and it is frequently inactivated in NK cell-derived lymphomas. Many target genes of PRDM1 have been reported; however, there is little insight into genes directly targeted by PRDM1 in NK cells. IL2 receptor signaling is crucial for many features of NK cell activation including sustained proliferation and survival. Here we evaluated whether CD25 (IL2Rα), a critical gene involved in NK cell activation, is transcriptionally repressed by PRDM1 in activated human NK cells.Methods: ChIP-Seq was applied on normal primary NK cells activated using a special co-culture system that involved genetically engineered NK cell target cells (i.e. K562-Cl9-mb21) to determine genomic locations occupied with PRDM1. PRDM1 binding site screen in CD25 was computationally performed using two different ChIP-Seq peak callers. Two malignant NK cell lines (NK92 and KHYG1) were transduced with a retroviral construct that co-expresses PRDM1α and GFP. CD25 transcription expression was evaluated with DNA microarray, RNA-Seq and RT-qPCR on GFP-sorted NK cells 48h post-transduction. CD25 expression was stably knocked-down with two different shRNAs in NK92 and KHYG1 cell lines. CD25 was ectopically expressed in primary human NK cells activated through coculturing with K562-Cl9-mb21. GFP competition assays were performed on CD25 shRNA-transduced NK cell lines or primary NK cells with ectopic CD25 expression in limiting IL2 concentrations, respectively.Results: PRDM1 occupancy on the 1st intron of CD25 was detected with CisGenome and MACS softwares in activated primary NK cells. Direct binding of PRDM1 on CD25 was further validated by ChIP-qPCR in normal activated NK cells. Two PRDM1α-transduced PRDM1-nonexpressing NK cell lines showed significant transcriptional repression of CD25 by DNA microarray and RNA-Seq, the repression of which was cross-validated with RT-qPCR. Stable knock-down of CD25 inhibited growth of two NK cell lines cultured in limiting IL2 concentrations. By contrast, ectopic expression of CD25 resulted in positive selection of ex vivo cultured primary NK cells.Conclusions: Altogether these results establish CD25 as a direct transcriptional target of PRDM1 in human NK cells. These observations provide additional support for the role of PRDM1 in termination of NK cell activation and growth, with implications on neoplastic transformation or NK cell function when it is deregulated. DisclosuresNo relevant conflicts of interest to declare.

CLIPick: a sensitive peak caller for expression-based deconvolution of HITS-CLIP signals

High-throughput sequencing of RNAs isolated by crosslinking immunoprecipitation (HITS-CLIP, also called CLIP-Seq) has been used to map global RNA–protein interactions. However, a critical caveat of HITS-CLIP results is that they contain non-linear background noise—different extent of non-specific interactions caused by individual transcript abundance—that has been inconsiderately normalized, resulting in sacrifice of sensitivity. To properly deconvolute RNA–protein interactions, we have implemented CLIPick, a flexible peak calling pipeline for analyzing HITS-CLIP data, which statistically determines the signal-to-noise ratio for each transcript based on the expression-dependent background simulation. Comprising of streamlined Python modules with an easy-to-use standalone graphical user interface, CLIPick robustly identifies significant peaks and quantitatively defines footprint regions within which RNA–protein interactions were occurred. CLIPick outperforms other peak callers in accuracy and sensitivity, selecting the largest number of peaks particularly in lowly expressed transcripts where such marginal signals are hard to discriminate. Specifically, the application of CLIPick to Argonaute (Ago) HITS-CLIP data were sensitive enough to uncover extended features of microRNA target sites, and these sites were experimentally validated. CLIPick enables to resolve critical interactions in a wide spectrum of transcript levels and extends the scope of HITS-CLIP analysis. CLIPick is available at: http://clip.korea.ac.kr/clipick/

Differential analysis of chromatin accessibility and histone modifications for predicting mouse developmental enhancers.

Enhancers are distal cis-regulatory elements that modulate gene expression. They are depleted of nucleosomes and enriched in specific histone modifications; thus, calling DNase-seq and histone mark ChIP-seq peaks can predict enhancers. We evaluated nine peak-calling algorithms for predicting enhancers validated by transgenic mouse assays. DNase and H3K27ac peaks were consistently more predictive than H3K4me1/2/3 and H3K9ac peaks. DFilter and Hotspot2 were the best DNase peak callers, while HOMER, MUSIC, MACS2, DFilter and F-seq were the best H3K27ac peak callers. We observed that the differential DNase or H3K27ac signals between two distant tissues increased the area under the precision-recall curve (PR-AUC) of DNase peaks by 17.5–166.7% and that of H3K27ac peaks by 7.1–22.2%. We further improved this differential signal method using multiple contrast tissues. Evaluated using a blind test, the differential H3K27ac signal method substantially improved PR-AUC from 0.48 to 0.75 for predicting heart enhancers. We further validated our approach using postnatal retina and cerebral cortex enhancers identified by massively parallel reporter assays, and observed improvements for both tissues. In summary, we compared nine peak callers and devised a superior method for predicting tissue-specific mouse developmental enhancers by reranking the called peaks.

New method for estimation of number of transcription factor binding sites using results of processing of ChIP-seq data by different peak callers

New method for estimation of number of transcription factor binding sites using results of processing of ChIP-seq data by different peak callers