clustering

Definition

Clustering in bioinformatics refers to computational methods that group biological entities (genes, proteins, samples, or cells) based on similarity patterns in their characteristics or behaviors. These unsupervised machine learning techniques identify natural groupings without prior labels, revealing functional modules, co-expressed gene sets, or disease subtypes. Common algorithms include hierarchical clustering, k-means, and graph-based methods like Markov clustering (MCL). Clustering is fundamental for analyzing high-throughput data such as RNA-seq, proteomics, and single-cell sequencing, helping researchers discover functional relationships, identify biomarkers, and understand biological organization. The quality of clustering depends on distance metrics, normalization methods, and algorithm parameters, making validation essential.

Visualize clustering in Nodes Bio

Nodes Bio enables researchers to visualize clustering results as network communities, where nodes represent biological entities and edges show similarity relationships. Users can apply community detection algorithms to identify functional modules within protein-protein interaction networks, visualize co-expression clusters in gene regulatory networks, or map cell type clusters from single-cell data onto molecular interaction networks to understand cell-state-specific pathway activity.

Example Use Case

A cancer researcher analyzing RNA-seq data from 200 tumor samples uses hierarchical clustering to identify four distinct patient subgroups based on gene expression patterns. By visualizing these clusters in Nodes Bio as a network where patients are connected by expression similarity, they discover that one subgroup shows elevated immune checkpoint gene expression. Mapping these genes onto protein interaction networks reveals a densely connected module involving PD-1, PD-L1, and CTLA-4, suggesting this patient subgroup may respond better to immunotherapy treatments.

Related Terms