Gaining biological insights through supervised data visualization

2026-06-30 · Source: Machine learning : nature.com subject feeds · Field: Science & Research — Life Sciences & Biology, Health & Medical Research, Mathematics & Computational Sciences · Depth: Expert, long

Summary

RF-PHATE is a novel supervised data visualization approach designed to interpret complex biological data by incorporating expert knowledge. Unlike unsupervised methods such as t-distributed stochastic neighbor embedding, Uniform Manifold Approximation and Projection, and Isomap, which often highlight irrelevant structures, RF-PHATE uses random forests to learn feature-label relationships. This process translates information into low-dimensional embeddings that reveal label-relevant structure while suppressing extraneous variation. The method handles large datasets and is suitable for both classification and regression tasks. Its utility was demonstrated across four case studies, including longitudinal multiple sclerosis data from MSBase, Raman spectral measurements of antioxidant effects, outcomes of patients with COVID-19, and RNA sequencing data with simulated dropout. These applications highlight RF-PHATE's ability to enhance interpretability and expose meaningful biological structures.

Key takeaway

For research scientists analyzing complex biological datasets, RF-PHATE offers a superior visualization tool to uncover label-relevant structures. You should consider integrating this supervised approach to move beyond the limitations of traditional unsupervised methods, ensuring your visualizations directly support downstream analysis goals and expert annotations. This can significantly improve data interpretability and accelerate biological discovery by focusing on meaningful variations.

Key insights

RF-PHATE integrates expert labels into dimensionality reduction for more relevant biological data visualization.

Principles

• Supervised visualization enhances interpretability over unsupervised methods.
• Random forests effectively learn feature-label relationships.
• Suppressing extraneous variation reveals meaningful data structure.

Method

RF-PHATE uses random forests to learn feature-label relationships, then translates this information into low-dimensional embeddings. This method is applicable for both classification and regression tasks.

In practice

• Analyze longitudinal multiple sclerosis patient data.
• Visualize Raman spectra for antioxidant effect studies.
• Interpret complex COVID-19 patient outcomes.

Topics

• Supervised Visualization
• Dimensionality Reduction
• Biological Data Analysis
• Random Forests
• Multiple Sclerosis
• RNA Sequencing

Code references

• zhd96/pi-vae
• jakerhodes/RF-PHATE

Best for: AI Scientist, Research Scientist, Data Scientist

Related on AIssential

• Explainable Iterative Data Visualisation Refinement via an LLM Agent — An LLM-driven agent automates and explains data visualization refinement by iteratively optimizing DR hyperparameters.
• The Art and Science of Data Visualization — Data visualization translates data into understandable insights through deliberate design decisions and visual encodings.
• StatQuest: Random Forests Part 2: Missing data and clustering — Random forests can impute missing data and cluster samples by leveraging leaf node co-occurrence for proximity.
• Cross-Domain Feature Expansion for Tabular Medical Data via Knowledge Graphs Injection — MedKGTab expands tabular medical features by combining data-driven statistics with biomedical knowledge graphs for high-fidelity generation.
• Multidimensional data analysis and classification using SMIAL — SMIAL simplifies label-free multidimensional image analysis for life scientists through a no-code, end-to-end GUI.
• A multimodal approach for visualizing and identifying electrophysiological cell types in vivo — PhysMAP integrates multiple electrophysiological modalities to accurately identify neuronal cell types and detect batch effects.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine learning : nature.com subject feeds.