Browsing large scale cheminformatics data with dimension reduction

Abstract

Visualization of large-scale high dimensional data tool is highly valuable for scientific discovery in many fields. We present PubChemBrowse, a customized visualization tool for cheminformatics research. It provides a novel 3D data point browser that displays complex properties of massive data on commodity clients. As in GIS browsers for Earth and Environment data, chemical compounds with similar properties are nearby in the browser. PubChemBrowse is built around in-house high performance parallel MDS (Multi-Dimensional Scaling) and GTM (Generative Topographic Mapping) services and supports fast interaction with an external property database. These properties can be overlaid on 3D mapped compound space or queried for individual points. We prototype use with Chem2Bio2RDF system using SPARQL query language to access over 20 publicly accessible bioinformatics databases. We describe our design and implementation of the integrated PubChemBrowse application and outline its use in drug discovery. The same core technologies can be used to develop similar high dimensional browsers in other scientific areas.

Stochastic Multidimensional Scaling

“…It is observed that MDS yields two distinct clusters of compounds, while the lower complexity compounds are scattered towards the edges. The visualization obtained here is comparable to those obtained in [8], [17].…”

Geo-Coordinated Parallel Coordinates (GCPC): A Case Study of Environmental Data Analysis

“…The main dilemma is how to represent high dimensional data in 2D display space without adding extra visual clutter. Several approaches have been introduced to visually represent high dimensional data, but only a few of them are scalable [5,18,33,55]. Additionally, challenges include identifying patterns with large a number of dimensions, and following trends across multiple dimensions.…”

“…The goal is to maintain the relations between the high dimension data in the lower dimension space without adding any new relations that are not originally in the data. Techniques such as principle component analysis (PCA) [72,55], multidimensional scaling (MDS) [3,33], and Self Organizing Maps (SOM) [3,57] are commonly used to transform data to lower dimensions.…”

“…For example, PCA computes the principal components from the original data, which could be easily used to classify and cluster the data points. Figure 2.4 shows an example of using MDS to visualize gene related chemical compounds in the PubChem database [33].…”

Emerging Computational Methods for the Life Sciences Workshop Special Issue

“…This collection of papers was selected from those presented at the Emerging Computational Methods for the Life Sciences Workshop of the ACM HPDC 2010 conference , Chicago, Illinois June 22, 2010. The papers were all enhanced over the conference versions and separately reviewed.…”