Transcription factors regulate gene activity by recognizing specific DNA sequences and influencing whether nearby genes are switched on or off. Their binding sites are embedded in chromatin, the complex of DNA and proteins that packages the genome. Because chromatin structure changes as cells develop or respond to signals, researchers need methods that can map both accessible regions and the physical protection provided by bound proteins. Chromatin-accessibility assays such as MNase-seq, DNase-seq and ATAC-seq infer this organization by fragmenting DNA with enzymes, sequencing the resulting pieces and examining where fragment ends occur. A bound transcription factor can shield a short stretch of DNA from cutting, leaving a footprint in the data. Yet the enzymes themselves may cut some DNA sequences more readily than others. Those sequence preferences can produce apparent footprints at sites where no transcription factor is present, complicating efforts to connect genomic patterns with regulatory activity.

V-plot analysis offers a way to visualize these fragment patterns in two dimensions. For each sequenced DNA fragment, researchers plot its length against the distance between its center and a selected transcription-factor motif. When many fragments are displayed together, their arrangement can form a V-like shape. Fragments that retain information about a protected region tend to occupy the area inside the V, while fragments lacking a remnant of the footprint are represented outside it. The shape is therefore not merely a one-dimensional dip in enzyme cutting; it reflects the relationship between fragment size and position relative to the motif. In routine analyses, researchers may identify a V-shape by eye, but the study argues that visual resemblance alone is insufficient. Sequence bias can generate a V-shaped distribution too, meaning that a motif-associated pattern must satisfy additional geometric and quantitative conditions before it is interpreted as evidence of transcription-factor occupancy.

To investigate the distinction, the researchers examined CTCF, a widely studied transcription factor, using low-dosage micrococcal nuclease sequencing in K562 cells. They compared three classes of genomic locations: highly occupied CTCF motifs positioned near an MNase sequence-bias site, weakly occupied CTCF motifs with a nearby bias sequence, and MNase-bias sequences lacking a nearby CTCF motif. V-shapes appeared in all three categories, confirming that the presence of the pattern itself could not establish binding. Their appearances, however, differed. At occupied CTCF motifs, the V-shape contained a visibly broad channel with few or no signals. This gap is consistent with a region protected from digestion by the bound factor. At MNase-bias sequences, the channel was much narrower, suggesting that the pattern could arise from preferential cleavage rather than a protected protein-DNA interface. Fragment signals were also enriched inside the V at occupied motifs, whereas bias-generated patterns showed a more even distribution between the inside and outside regions.

The authors used several complementary observations to test that interpretation. A CTCF chromatin immunoprecipitation dataset showed strong enrichment within the inside-V area at occupied motifs, an independent indication that the pattern coincided with CTCF binding. The same enrichment was not seen at MNase-bias sequences. Analysis of fragment ends revealed another difference: at occupied motifs, the two flanking digestion hotspots were separated by a broad interval, consistent with a protected footprint between them. That separation was absent at bias sequences. Similar graphical distinctions emerged when the researchers analyzed DNase-seq and ATAC-seq data, despite the different enzymes and experimental principles involved. DNase I and the Tn5 transposase used in ATAC-seq also have sequence preferences, so the recurring contrast suggests that the proposed tests may help identify enzyme-generated patterns across multiple accessibility platforms rather than being limited to MNase-based experiments.

The first criterion is the width of the V-channel, the horizontal region around the motif that is depleted of signals. In principle, this width should approximate the minimum footprint size—the portion of DNA stringently protected under the experimental conditions. The second is inside-V enrichment, which compares fragment density in the interior of the V with density outside it. The researchers used the ratio of these areas to quantify how strongly fragments were concentrated within the footprint-associated region. Their experiments with MAZ, another transcription factor, showed that inside-V enrichment tracked the factor’s occupancy level: motifs with greater enrichment also displayed stronger MAZ chromatin immunoprecipitation signals. The channel width, by contrast, remained relatively constant across groups with different occupancy levels. This separation suggests that enrichment can reflect how much a factor is present, while channel width may reflect an intrinsic aspect of its DNA contact or the way the experiment captures protection.

These measurements are not fixed universal values. The study found that different transcription factors produced different channel widths, potentially because their structures contact DNA through distinct interfaces. Even for the same factor, channel width varied between datasets, indicating that enzyme kinetics and other experimental conditions influence the apparent length of a footprint. Inside-V enrichment was comparatively stable for a given factor across the datasets examined, although the authors emphasize that both parameters are needed for robust classification. They developed a statistical pipeline that uses channel width and inside-V enrichment to separate V-shapes at transcription-factor motifs from those at bias sequences in MNase-seq, DNase-seq and ATAC-seq data. The approach also helped distinguish occupied CTCF motifs from shuffled or unoccupied motifs, with inside-V enrichment alone providing useful separation when long candidate bias sequences were present.

The findings do not eliminate the need for independent validation, and they do not imply that every footprint can be interpreted without considering enzyme bias, motif quality or cellular context. Instead, they provide a practical framework for making V-plot evidence more discriminating, particularly when high-quality chromatin immunoprecipitation data are unavailable. The researchers also reproduced key features of occupied CTCF footprints through in silico MNase digestion, supporting the proposed relationship between protected DNA and the observed geometry. The study’s sequencing data are available through the Genome Sequence Archive under accession HRA017804, and its analysis code is available on GitHub. By requiring a sufficiently broad protected channel together with measurable fragment enrichment inside the V, the authors conclude that researchers can reduce false discoveries while retaining a straightforward way to inspect transcription-factor binding across chromatin-accessibility datasets.

A useful implication of the study is that footprint discovery should be treated as a comparative classification problem rather than as a search for an isolated visual pattern. The relevant question is not simply whether a V-shape occurs near a recognition motif, but whether its geometry and fragment distribution resemble the behavior expected from protected DNA more closely than the behavior of a sequence favored by the assay enzyme. This distinction is especially important at genomic positions where a transcription-factor motif lies close to a cleavage-bias sequence, because the two signals can overlap and make a one-dimensional accessibility profile difficult to interpret.

The proposed measurements also separate two biological or technical features that are often conflated. Inside-V enrichment provides information about the concentration of footprint-bearing fragments and, in the MAZ analysis, changed with the apparent level of occupancy. Channel width instead describes the span of the protected interval under the particular digestion or transposition conditions. Thus, a weak footprint should not automatically be interpreted as a smaller protein-DNA interface, and a broad channel should not automatically be interpreted as stronger binding. The two quantities answer different questions and should be evaluated together when assigning confidence to a candidate site.

This framework offers a way to organize controls within an analysis. Motif-centered plots can be compared with plots centered on bias sequences, shuffled motifs or motifs considered unoccupied, while keeping the sequencing assay and processing strategy consistent. Such comparisons help reveal whether a threshold separates biological signal from an assay-specific background. The study’s examples further suggest that classification cutoffs cannot be assumed to transfer unchanged between transcription factors or between MNase-seq, DNase-seq and ATAC-seq experiments. A robust application therefore requires estimating the relevant distributions in the dataset being analyzed, rather than importing a single footprint size or enrichment value from another experiment.

The in silico digestion results provide a mechanistic bridge between the diagrams and the sequencing observations. When cleavage is restricted across a motif-sized interval, the simulated fragments reproduce the broad vacant channel and the concentration of fragments within the inside-V region seen at occupied sites. This supports the interpretation that the geometry can arise from protection of DNA rather than from an arbitrary plotting artifact. At the same time, the study’s comparison with CTCF chromatin immunoprecipitation illustrates why orthogonal evidence remains valuable: V-plot criteria improve the specificity of inference, but they do not replace information about protein occupancy when such information can be obtained.

For researchers applying the method, the practical output is therefore more informative than a binary label alone. A candidate motif can be accompanied by its channel-width measurement, inside-to-outside fragment ratio and comparison with appropriate bias controls. Reporting these features makes it easier to distinguish sites supported by both criteria from sites driven mainly by one unusual signal. It also creates a clearer basis for comparing occupancy patterns across motifs, cell states or assay types, while preserving the simple visual inspection that makes V-plots attractive for genome-wide chromatin analysis.

Subject of Research: Transcription-factor footprint detection using V-plot analysis

Article Title: Two essential criteria for transcription factor footprint discovery using V-plot

Article References: Two essential criteria for transcription factor footprint discovery using V-plot. (n.d.). https://doi.org/10.1038/s44320-026-00242-5

Image Credits: AI Generated

DOI: 10.1038/s44320-026-00242-5

Keywords: transcription factors, V-plot analysis, chromatin accessibility, MNase-seq, DNase-seq, ATAC-seq, enzyme sequence bias, DNA footprints, essential, criteria, transcription, factor