liveBook · Manning

Unified distance matrix

As explained in Build AI Drug Discovery Pipelines

A unified distance matrix is a self-organizing-map visualization that represents the average distances between each node and its neighboring nodes.

A unified distance matrix (UMAT) visualizes a self-organizing map by coloring each node according to its average distance from neighboring nodes. Lighter regions indicate nearby, similar nodes and suggest clusters that preserve local structure from high-dimensional molecular fingerprints. section 7.1.2

Use the UMAT to:

  • Locate reliable clusters across lighter regions of the map. section 7.1.2
  • Overlay connected components to make cluster boundaries visible and identify compounds from separate nodes that may share features. section 7.1.2
  • Compare neighboring regions containing active and inactive compounds, including areas where small structural changes may correspond to activity changes. section 7.1.2

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Defines neighbor-distance coloring and extends it with connected components, stopping short of deriving the full SOM training procedure.
Examples
Uses a UMAT with active/inactive compound enrichment and connected components linking positions such as (0, 0) and (0, 4).
What it assumes
Assumes familiarity with self-organizing maps and high-dimensional molecular fingerprints while motivating local-structure preservation.
What you can do after
Apply the visualization to identify clusters and investigate structure–activity changes across neighboring map regions.
How it connects
Links self-organizing maps, molecular fingerprints, local structure, clustering, and drug repurposing.

This explainer was generated by Manning's knowledge graph from the text of Build AI Drug Discovery Pipelines. The section links point to the parts of the book it draws on.