A unified framework for the integration of multiple hierarchical clusterings or networks from multi-source data

Area of Science:

• Computational Biology
• Bioinformatics
• Data Integration

Background:

• Integrating heterogeneous data is a key challenge in computational biology.
• Existing methods primarily focus on integrating similar data types (e.g., numerical tables).
• Data often exists in diverse formats like trees, networks, and factorial maps, hindering integrated analysis.

Purpose of the Study:

• To develop a procedure for integrating and comparing heterogeneous data representations, specifically trees and networks.
• To provide a computational framework for analyzing complex biological datasets with varied structures.

Main Methods:

• A two-step procedure involving projection into a common coordinate system followed by multi-table integration.
• Projection is achieved using distance matrices and multidimensional scaling.
• Integration utilizes multiple factor analysis on the projected data coordinates.

Main Results:

• The method successfully integrates and compares data from diverse sources, including tree and network structures.
• The approach is validated on simulated data and applied to real-world biological datasets.
• It offers an alternative to traditional kernel methods for data integration.

Conclusions:

• The developed procedure effectively integrates and compares heterogeneous data representations.
• The method is applicable to biological data, such as single-cell transcriptomics and cancer network analysis.
• This approach enhances the ability to study patterns and interactions within complex biological systems.