Networks and Graphs in Biological Data: Current Methods, Opportunities and Challenges

This article collection showcases innovative computational and analytical approaches to understanding complex disease mechanisms, with a particular focus on cancer genomics and COVID-19 outcomes. The first article demonstrates the power of network-based methods, such as DiWANN and bipartite network analyses, for identifying driving gene mutations and understanding their role in cancer heterogeneity and potential therapeutic targets. The second article introduces a novel graph-based visualization framework for comparative genomic sequence alignments, maintaining sequence order context and enhancing structural variation analysis through hierarchical layouts and dedicated algorithms. The third article explores the integration of machine learning and deep learning techniques—including advanced QSAR modeling—to optimize combinational drug therapies in breast cancer, highlighting how deep neural networks can effectively generalize structure-activity relationships across diverse compounds. The fourth article applies Bayesian network modeling to reveal how specific body fat distributions, especially elevated visceral and liver fat, influence hospitalization risk in COVID-19, surpassing traditional BMI assessments. Collectively, these articles emphasize the pivotal role of computational strategies—spanning network analysis, graph visualization, and machine learning—in elucidating disease patterns, predicting clinical outcomes, and informing personalized medicine.
----

This article collection highlights cutting-edge computational and analytical approaches in biomedical research, focusing on cancer genomics, breast cancer therapeutics, comparative genomics, and the relationship between body fat distribution and COVID-19 hospitalisation risk. The first article demonstrates how network-based models can identify driver gene patterns across tumor types, facilitating the detection of both common and cancer-specific mutations while reducing computational costs. The second article introduces a graph-based visualization framework for comparing genome structures, providing essential insights into structural genomic variation that are often lost in traditional sequence alignments. The third article explores the integration of machine learning and deep learning models to predict the synergistic effects of combinational drug therapies in breast cancer, with deep neural networks showing high predictive accuracy in modeling structure-activity relationships. Finally, the fourth article employs Bayesian network modeling to reveal how visceral and liver fat play a critical role in determining hospitalisation outcomes for COVID-19 patients, underscoring the centrality of fat distribution over BMI in risk assessment. Collectively, these studies showcase how computational innovations, from network analysis to deep learning and probabilistic modeling, are advancing our understanding of disease mechanisms, therapeutic strategies, and health risk factors.
----

Networks are ever-present in biology. Whether in the interplay of factors influencing gene expression, the balance of consumers and producers in food webs, or the interactions between parts of an enzyme that modulate its activity, hardly anything in biology acts completely independently. Understanding the interdependencies is often critical to understanding the behavior of a system. Unfortunately, few tools exist for computing or visualizing network data that utilize many features common in biological networks. At the same time, there are properties of biological networks that cannot be well-captured by current graphs and network formalisms. This Research Topic will bring together descriptions of the current best practices for visualizing and computing the breath of network types present in biological data, new tools that present unique capabilities for this field, and concise descriptions of outstanding challenges in correctly representing real biological networks as the formal graph/network structures.

This Research Topic intends to provide an analysis of the landscape of network approaches applied to biological data and serve as a sourcebook of best practices for dealing with archetypal data types. It will introduce new tools uniquely suited for the challenges of biological network data and will bring together illustrations of remaining problems (as well as corresponding data sets) in biological networks that are not yet addressed. In addition to other poorly served areas of biological network representations, it will highlight the challenges and potential opportunities that exist in situations such as biological networks that have "natural" node orderings or positionings that are independent of edge weights, networks with conditionally-valued edge weights, and metagraph and hypergraph features within biological networks. By highlighting both the current best practices for biological end-users, and areas where current tools could be improved, this Research Topic aims to improve the utilization of network visualization and analytics in the biological sciences, bring potential areas for new contributions to the attention of the broader graph/network community, and invite improvements from areas such as transportation network graphs that have useful but as-yet unexploited applications to biological data.

This Research Topic accepts - but is not limited to - submissions in any aspect of network or graph visualization or visual-analytic approaches to biological data, such as:

• Reviews detailing current best-practices for specific biological domains for either Exploratory or Explanatory purposes.
• Methods papers introducing new approaches or software tuned to specific biological network-visualization needs in either Exploration or Explanation.
• Research papers comparing or contrasting the performance of different network visualization/analytics approaches to specific biological domain questions and data.
• Position papers illustrating challenges in biological network visualization that are not well-served by current approaches, with corresponding demonstration data sets.

Keywords: Networks, Graphs, Interactions, Dependencies, Non-Linear, Exploration, Analysis, Computational biology, Personalized medicine, QSAR modeling, Deep learning, Machine learning, Graph visualization, Network analysis

Important note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.