Eigen RadarScience
Analysis

A branching cell map tracks lineage; an automated sinus score catches treatment change

Bonsai maps single-cell data as a distance-preserving tree and identified a new immune-cell subtype. An automated sinus score also tracked post-treatment change more responsively than visual scoring.

Science··Evening
Translucent branches extend from blue and violet cells in the left foreground; a connected offshoot separates into coral-red cells across a luminous ivory-and-aqua microscopic volume.

Bonsai preserves cell proximity along branches

Developed at the University of Basel, Bonsai draws single-cell RNA sequencing data as a branching tree instead of compressing it onto a flat map. Distances measured along the branches preserve how closely cells sit in the high-dimensional data, allowing developmental paths and cell types to remain visible in one structure. Known relationships among human blood cells reappeared in the tree. The software is freely available to researchers, so lineage relationships in other single-cell datasets can be examined with the same tree-shaped representation.[1]

The tree found an unexpected immune-cell lineage

When Bonsai was applied to human blood cells, it recovered known cell-type relationships and distinguished a previously undescribed natural killer-cell subtype. Those cells arose from the myeloid lineage rather than the lymphoid lineage expected for natural killer cells. First author Daan de Groot says distance along the branches faithfully reflects proximity in the original data space. The new subtype became a concrete example of the tree doing more than arranging familiar clusters: the team reports that it also made a different developmental path visible.[1]

An automated sinus score followed treatment change more closely

A deep-learning sinus-severity score from the National Jewish Health team tracked change after treatment more responsively than the widely used Lund-Mackay visual score. CT imaging reveals what is happening inside the sinuses, yet traditional scoring can remain dependent on the reader's judgement. The automated method aims to turn change in the same images into a more consistent number. Senior author Stephen M. Humphries says CT opens a view of disease inside the sinuses while traditional scoring can remain subjective.[2]

References

  1. News sourcePhys.orgSoftware that draws cell data as a branching tree instead of squashing it onto a map↩1↩2
  2. News sourceMedical XpressAn automated score for sinus CT tracked change after treatment better than the eye↩