Dimension reduction of clustering results in bioinformatics

TL;DR

This paper presents a visualization method for density-based clustering results, specifically OPTICS, that enhances interpretability and aids in identifying biologically relevant clusters in high-dimensional bioinformatics data.

Contribution

The work introduces a novel visualization approach for OPTICS clustering results that incorporates hierarchical prior knowledge to improve cluster identification in bioinformatics.

Findings

Effective visualization of clustering structure in high-dimensional data

Identification of biologically relevant clusters using the proposed method

Application demonstrated on two bioinformatics datasets

Abstract

OPTICS is a density-based clustering algorithm that performs well in a wide variety of applications. For a set of input objects, the algorithm creates a so-called reachability plot that can be either used to produce cluster membership assignments, or interpreted itself as an expressive two-dimensional representation of the density-based clustering structure of the input set, even if the input set is embedded in higher dimensions. The main focus of this work is a visualization method that can be used to assign colours to all entries of the input database, based on hierarchically represented a-priori knowledge available for each of these objects. Based on two different, bioinformatics-related applications we illustrate how the proposed method can be efficiently used to identify clusters with proven real-life relevance.