Context-situated visualization of biclusters to aid decisions: Going beyond subspaces with parallel coordinates

Daniel Gonçalves, Rafael S. Costa, Rui Henriques

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Pattern discovery and subspace clustering are pervasive tasks across biological, biotechnological, and biomedical domains. Parallel coordinates plots and heatmaps are reference visualizations for individual biclusters. Both have been object of improvements over time, with a special emphasis on heatmaps, commonly used in gene expression analysis. However, the emphasis is solely placed on the corresponding subspace, preventing an assessment of biclusters' significance against global regularities. This work proposes an improvement on bicluster visualization by disruptively extending parallel coordinates representations with the means to compare the local bicluster against the remaining dataset instances helping in the contextualization of a pattern in the broader picture of an entire dataset. The proposed solution is the first able to deal with mixed data types and is independent from the underlying biclustering or pattern mining algorithm. Results in different data domains show the utility of the proposed visualization, especially in primary phases where visual inspection of biclusters is used.

Original languageEnglish
Title of host publicationAVI 2022
Subtitle of host publicationProceedings of the Working Conference on Advanced Visual Interfaces
EditorsP. Bottoni, E. Panizzi
Place of PublicationNew York
PublisherACM - Association for Computing Machinery
Number of pages5
ISBN (Print)978-145039719-3
DOIs
Publication statusPublished - Jun 2022
Event16th International Conference on Advanced Visual Interfaces - Frascati, Roma, Italy
Duration: 6 Jun 202210 Jun 2022

Publication series

NameProceeding of Advanced Visual Interfaces (AVI)
PublisherAssociation for Computing Machinery

Conference

Conference16th International Conference on Advanced Visual Interfaces
Country/TerritoryItaly
CityRoma
Period6/06/2210/06/22

Keywords

  • biclustering
  • bioinformatics tool
  • data science
  • omics/clinical data visualization
  • subspace analysis

Fingerprint

Dive into the research topics of 'Context-situated visualization of biclusters to aid decisions: Going beyond subspaces with parallel coordinates'. Together they form a unique fingerprint.

Cite this