Colocalization

The Colocalization command identifies objects from two or more input classes that overlap each other. Overlapping objects are defined as colocalising. EVAnalyzer assigns all colocalising objects a shared tracking ID, allowing them to be grouped in the results table.

Unlike Object Math, which derives a new shape from two object sets, Colocalization’s job is to link objects that already exist - each overlapping pair (or group) keeps its own identity and metrics, but gains a shared tracking ID and, optionally, a new object recording the overlap area itself.

Parameters

Parameter	Description
Classes to colocalize	Two or more object classes that must all have an overlapping member for an object to be considered colocalising
Multiplicity	How many partners an object may coloc with at once - see below
Class for overlapping areas	A class assigned to a new object representing the actual overlap area
Exclude classes	Objects that also overlap one of these classes are dropped entirely, even if they’d otherwise match - see below
Min colocalization area	Minimum overlap area in the selected unit
Size unit	Unit for the min area threshold

Note

Filter classes, an older secondary-overlap filter, still exists in the underlying project schema for compatibility with older projects, but is hidden from the command picker in favor of Exclude classes below.

Multiplicity

Multiplicity replaces the old “Allow multi-object colocalization” checkbox with three explicit modes:

Mode (GUI label)	Effect
No multi coloc (1:1) (default)	Every object keeps only its single best-overlap match (largest overlap area) per other class
Allow multi coloc	Every object may coloc with every overlapping partner that meets Min colocalization area
Multi coloc only for selected	Only the listed classes may coloc with more than one partner - every other class in Classes to colocalize is still capped to its single best match

Multi coloc only for selected is the tool for a “count spots per cell” shape: with Classes to colocalize set to Cell, Spot and multiplicity limited to Cell, a cell can coloc with any number of spots, but each spot still picks exactly one cell (the one it overlaps most) - so a spot sitting on the border between two cells is never double-counted under both.

Where an object is capped to a single best match, ties (equal overlap area) are broken toward the lower object ID, so the result is deterministic rather than depending on cache iteration order.

Exclude Classes

Exclude classes is a blocklist evaluated after the classes-to-colocalize check: an object that satisfies Classes to colocalize is dropped entirely - no colocalization recorded, no overlap-area object created - if it also overlaps any object from one of these classes by at least Min colocalization area. Leave empty (the default) to disable this filter.

Example: to find objects colocalizing with a nucleus stain and a membrane stain, but excluding any that also overlap a dead-cell marker, set Classes to colocalize to the nucleus and membrane classes and Exclude classes to the dead-cell-marker class.

Colocalizing handling

When an object is found to colocalize, one of two operations is applied:

• Move - the object’s class is changed to a new class. The Object ID is unchanged.
• Copy - a new object is created with the new class; the origin object’s ID is recorded in the copy’s origin object ID field.

A new colocalizing area object is created for the overlapping region. This object is assigned to the class specified by Class for overlapping areas and can be used for further measurements.

All colocalising objects - input objects and the colocalization-area object - receive the same tracking ID. In the results table, rows with the same tracking ID appear adjacent to each other.

Example: Two-channel spot colocalization

1. Classify spots in Cy5 channel → cy5@spot
2. Classify spots in Cy7 channel → cy7@spot
3. Add a Colocalization step:
   • Classes: cy5@spot, cy7@spot
   • Class for overlapping areas: coloc@cy5cy7
4. All Cy5 spots that overlap with a Cy7 spot (and vice versa) receive a common tracking ID.