NBBC: a non-B DNA burden explorer in cancer

Abstract

Alternate (non-B) DNA-forming structures, such as Z-DNA, G-quadruplex, triplex have demonstrated a potential role in cancer etiology. It has been found that non-B DNA-forming sequences can stimulate genetic instability in human cancer genomes, implicating them in the development of cancer and other genetic diseases. While there exist several non-B prediction tools and databases, they lack the ability to both analyze and visualize non-B data within a cancer context. Herein, we introduce NBBC, a non-B DNA burden explorer in cancer, that offers analyses and visualizations for non-B DNA forming motifs. To do so, we introduce 'non-B burden' as a metric to summarize the prevalence of non-B DNA motifs at the gene-, signature- and genomic site-levels. Using our non-B burden metric, we developed two analyses modules within a cancer context to assist in exploring both gene- and motif-level non-B type heterogeneity among gene signatures. NBBC is designed to serve as a new analysis and visualization platform for the exploration of non-B DNA, guided by non-B burden as a novel marker.

Graphical Abstract

Figure 1.

The overall design of NBBC. (A) The input includes genomic regions in query and non-B types. (B) The first module is ‘Gene Screen’. The gene layer analyses non-B burden for input gene query. And the second module is ‘Motif screen’ that performs sequence clustering. (C) The output includes the dissection of burden in the query regions, genes with high burdens and non-B DNA sites with user-desired features.

Figure 2.

Introduction to input options. NBBC supports multi-level burden query and the current version provides four options at three levels for different goals and user circumstances. (A) Signature-level input. The input includes popular cancer signatures, cell line molecular signatures or user-defined signatures. (B) Gene-level input. The typical use case is a quick single gene search by manual input and motif exploration with the query gene. (C) Site-level input. It applies to burden queries at the high genomic resolution, such as cancer-specific mutation sites or regions with copy number alterations.

Figure 3.

Gene screen layer module to explore non-B burden. The gene layer analyzes non-B burdens and provides several visualizations for descriptive analysis of burden values, burden distribution, and burden-based gene clustering. (A) A stacked bar plot is used to visualize the total non-B burden. (B) The distribution of non-B burdens by non-B structure motif types. (C) A bubble plot allows users to observe the non-B burden by gene and type. (D) A burden clustering function is also available in the heatmap format.

Figure 4.

Motif screen layer module to explore viable non-B forming sequences. We use this module to further select high-quality motifs that are more likely to form non-B structures in the interested genes and to provide the specific sequences that can be used for wet lab experiments. (A) The interface for motif clustering. Users can choose two of pre-summarized motif sequence features to perform 2D clustering. (B, C) Clusters of motifs based on sequence features. In the example, motifs with high Guanine content (G%) and decent lengths are highlight as good candidates for user to focus on. Two visualizations are presented labeling gene names (left, gene-informed) or non-B types (right, non-B informed). (D) For cases where users are interested in flank regions of a motifs, the app helps output left- and right-flank regions for motifs based on user-input length. The step is achieved with real-time reference genome query.