Co-expression Analysis

Co-expression analysis compares two stainings together instead of interpreting each staining separately.

The correct interpretation depends on how cells were detected.

DAPI or nuclei-source workflows

In cell counting and classification projects, both marker classifiers use the same shared cell population from the Cell Detection marker.

That means co-expression can compare marker-positive and marker-negative states for the same detected cell IDs.

Common modes include:

• positive / positive: cells expressing both markers
• positive / negative: cells expressing the first marker but not the second
• negative / positive: cells expressing the second marker but not the first
• negative / negative: cells expressing neither marker
• either positive: cells expressing one marker or both

Independent-detection workflows

In independent-detection projects, each marker has its own detections. Cell IDs are not shared across stainings.

NeuroQP therefore uses mask-overlap matching to compare marker detections. It checks whether accepted marker-positive detections from two stainings overlap enough to be treated as the same biological cell or object.

For these workflows, co-expression should be interpreted as marker-positive overlap between independently detected populations.

Useful modes include:

• marker A positive overlapping marker B positive
• marker A positive without an overlapping marker B positive cell
• marker B positive without an overlapping marker A positive cell
• either marker-positive population, counting overlaps once

Negative-negative co-expression is not meaningful for independent-detection projects because negative means rejected candidate, not a biological negative class.

Before you calculate

Co-expression is only meaningful if both stainings are individually trustworthy first.

Before using it, make sure:

• both selected models look reliable in Results
• registration is good enough in the regions being compared
• the comparison matches a real biological question

Add co-expression to a dashboard plot

1. Review each staining individually in Results and confirm its selected classifier.
2. On desktop, open Statistics, choose Edit plots, and edit an existing plot or add one.
3. In the staining selector, choose a compatible pair and its co-expression mode. Positive/positive is a useful starting point.
4. Include the individual stainings as additional sources, or keep their summaries in neighboring dashboard plots.
5. Choose a valid metric, grouping and filters, then Save.
6. Use Process statistics or Reprocess statistics if the required pair results are missing or outdated.

For example, compare cFos-positive cells, TRAP-positive cells and cFos/TRAP positive/positive cells across the same groups and regions. Keep the metric and filters consistent when comparing these populations.

The order of the pair matters for asymmetric modes. A positive/B negative describes a different population from A negative/B positive. Use these modes only when they answer a specific population question.

A plot can contain several individual stainings and pair outputs. Their selection and order are saved with that plot. Classifier choices remain personal, so confirm them when comparing the shared dashboard with a colleague.

See Using the statistics dashboard for editing and viewing controls, and Metric semantics by workflow for valid counts, proportions and densities.