Graphs have been successfully used in several branches of science to visualize complex relationships, interactions, outcomes of complicated processes and even to simplify scientific notation. Graph theory has been largely applied to multiple scientific disciplines. For example, promoting the development of Chemical Graph Theory (CGT) allowed the combinatorial and topological exploration of the chemical molecular structure as well as its numerical description.

In recent years, graph-based deep learning has emerged as a powerful tool for modeling complex systems in both cheminformatics and bioinformatics. Knowledge graph-based deep learning, and by extension, any NLP or LLM-based approaches used to build such knowledge graphs, have played a significant role in advancing research in fields such as drug-target associations. Methods such as graph-attention (GAT) networks and graph convolutional networks (GCNs) have been widely used for tasks such as activity/function prediction and interaction modeling.

Graphs have been applied in cheminformatics for modeling drug design and drug-receptor relationships, thus having a direct impact on medicinal chemistry. With the emergence of the omics sciences, graph-based methods have been extended to bioinformatics through the graphical characterizations of DNA/RNA and protein structures for comparative sequence analysis without the use of alignment algorithms. On the other hand, complex biological systems such as metabolic, transcriptional regulatory and protein interaction networks were also simplified into the topology of graphs/networks to gain useful insights into such systems.

Currently, heterogeneous graphs are offering a solution to integrate several omics data types with the common intervention of chemical data such as in drug-protein interaction networks; it misses a compilation on the progressive use of graphs from cheminformatics to bioinformatics, arriving at the integration of chemical and biological data in heterogeneous graphs and at the synergies between graph-based modeling and machine learning algorithms in both fields.

This Research Topic accepts studies considering:

(i) Graph-based representations for both organic molecules and biomacromolecules as well as their usefulness to perform downstream analyses e.g., their numerical characterization, pairwise similarity analyses, clustering, activity/function prediction and interaction modeling.

(ii) Graphs representing complex biological/molecular relationships, e.g., heterogeneous graphs integrating omics and chemical data.

(iii) Application of graph neural networks, including GAT and GCN, to graph-modeled molecular and biological scenarios for deep learning tasks.

(iv) Graph database technologies applied to integrate biological/chemical/medical information.

Articles can be submitted as:

• Reviews detailing the state-of-the-art innovations from the aforementioned subjects.

• Methods papers introducing graph-based approaches or software aimed to improve current protocols or fill gaps in research.

• Research papers applying graphs/networks to provide new insights for analyzing chemical or biological systems. Comparing their performance with existing methodologies.

• Position papers illustrating current challenges and perspectives from graph-based approaches when applied to chemo-, bio-, and medical informatics. Critiques in current benchmarks and suggestions for better benchmarks.

Submit your idea

Impact

Frontiers in Bioinformatics

Submit your idea

Impact