Quantile map: simultaneous visualization of patterns in many distributions with application to tandem mass...

Area of Science:

• Biomedical research
• Data mining
• Bioinformatics

Background:

• High-throughput experiments generate high-dimensional data, posing challenges for analysis.
• Effective data reduction, summarization, and visualization are crucial for initial data exploration.

Purpose of the Study:

• Introduce a novel visualization tool, the quantile map.
• Demonstrate its effectiveness in analyzing high-throughput data, specifically tandem mass spectrometry data.

Main Methods:

• The quantile map visualizes probabilistic distributions using concentric doughnuts with gradient colors.
• Doughnut width is proportional to Fisher information for unbiased visualization.
• A parametric empirical Bayes (PEB) approach improves Fisher information estimation over maximum likelihood estimate (MLE).
• Hierarchical clustering and Hue-Chroma-Luminance color models enhance visualization in 2D grids.

Main Results:

• The quantile map effectively presents information from probabilistic distributions.
• PEB approach shows improvement over MLE for Fisher information estimation.
• Enhanced visualization techniques improve the display of multiple distributions.

Conclusions:

• The quantile map is a superior tool for summarizing and visualizing high-throughput data.
• This method aids in the initial exploration and analysis of complex biomedical datasets.