This page enables integrated analysis of regulatory effects for user-defined genetic variants by combining chromatin accessibility signals, deep-learning–based regulatory effect prediction, GWAS association evidence, and local gene structure annotation within a unified genomic context.

Manual Variant Input: Users may manually input one or more variant identifiers (e.g. vg-format IDs). Variants are parsed by chromosome and genomic position and matched to the corresponding reference genome coordinates.

File Upload: Variant sets can also be uploaded in standard formats, including:

• VCF files, containing chromosomal positions and reference/alternative alleles
• GWAS summary tables, containing CHR, POS, A1, A2, and P-value columns

Uploaded variants are automatically validated, filtered, and matched against the internal variant database. Only variants with consistent genomic coordinates and allele definitions are retained for downstream regulatory analysis.

Variant Filtering and Validation: To ensure reliable regulatory interpretation, input variants are subject to multiple quality and consistency checks, including:

• All variants must be located on a single chromosome
• The total number of variants must not exceed predefined limits
• The genomic span of variants must fall within a defined window
• Alleles from uploaded files are matched against database reference alleles

Deep Learning–based Regulatory Scoring: For each validated variant, regulatory effect scores are retrieved from deep learning models trained to predict chromatin accessibility changes from DNA sequence. In the current implementation, scores are derived from Basenji model (Kelley et al., 2018) , trained on large-scale ATAC-seq datasets.

Regulatory effect scores quantify the predicted change in chromatin accessibility caused by the alternative allele relative to the reference allele. Positive scores indicate an increase in predicted accessibility, whereas negative scores indicate a decrease. These scores provide a quantitative measure of the potential regulatory impact of non-coding variants.

Tissue-specific Effects: Regulatory effect predictions are provided across a wide range of tissues and developmental stages. This allows direct comparison of tissue-specific regulatory impacts for the same variant, facilitating identification of context-dependent or pleiotropic regulatory effects.

Regulatory Effect Heatmap: Tissue-specific regulatory scores are summarized using a heatmap representation, in which rows correspond to variants and columns correspond to tissues. Color intensity reflects the magnitude and direction of predicted regulatory effects, enabling rapid visual identification of variants with strong, tissue-restricted, or broadly acting regulatory consequences.

Chromatin Accessibility Landscape: Chromatin accessibility profiles derived from ATAC-seq experiments are displayed across the genomic region surrounding the queried variants. Accessibility signals are shown for multiple tissues, providing baseline chromatin context against which predicted variant-induced regulatory changes can be interpreted.

GWAS Integration: When GWAS summary statistics are provided, variants are annotated with their association significance and integrated with regulatory effect predictions. Scatter plots relate predicted regulatory effect scores to GWAS association strength (e.g. –log₁₀(P)), facilitating interpretation of potential regulatory mechanisms underlying complex trait associations.

Gene Context Annotation: Gene models overlapping the queried genomic interval are displayed alongside variant and regulatory information. This provides structural context for interpreting regulatory effects relative to gene bodies, exons, introns, and transcriptional orientation.