Once a niche innovation, spatial biology is now mainstream in both discovery and translational research. Spatial proteomic and transcriptomic technologies from Akoya Biosciences, Inc., Standard BioTools, 10x Genomics, and Bruker Spatial Biology have become widely accepted across research and Pharma.
In the early days, the most common question we got was – “Can you help us capture spatial data?”. Today, the question has shifted to something more exciting – “Now that we have spatial data… what do we do with it?”
Here are three key insights from those conversations:
A common starting point is to look at cell compositions and neighborhood changes between cohorts. That’s a good baseline — but what’s next?
We’re learning that biological meaning often hides in spatial context. For instance, in some indications, CD8+ T cell exhaustion shifts based on proximity to tumors. We’ve developed ways to quantify this — like spatial exhaustion scores, immune-tumor proximity scores, and a newer proximity-linked exhaustion score.
As an example, check out Sizun Jiang’s paper (Yeo et al. 2024) on EBV-linked classical Hodgkin lymphoma (cHL) — it shows how proximity of CD8+ T cells to tumor rich regions changes their exhaustion status in EBV+ cHL.
We have seen this time and again: data quality makes or breaks a project. Issues like inconsistent staining, poor segmentation, or batch variability can disrupt downstream workflows — but they’re manageable with the right quality controls in place.
As Sizun Jiang often says, “no assay is perfect,” which means success comes down to understanding the limitations of an assay and designing around them.
In practice, quality control can take longer than analysis itself. That’s why at Elucidate Bio, we’re developing models to automate and scale QC, so teams can move faster with more confidence — and focus their energy on generating insights, not cleaning up inputs.
Once your spatial data is high quality and well-characterized, the next step is making it actionable.
Imagine a system where you input spatial data and get back real insights: predicting patient outcomes, identifying responders, selecting the best drug combinations. That’s the vision of precision medicine — and AI makes it possible.
The key is using models built to understand spatial context and biological complexity. KRONOS, the spatial proteomics foundation model from Sizun Jiang and Faisal Mahmood, is a great example (Shaban et al. 2025). It achieved high predictive accuracy (ROC-AUC 0.78) — in a ccRCC dataset of just 27 patients.
Spatial Biology is where Genomics was 15 years ago — and we get to build the field with real-world impact in mind: for patients, for pharma, and for scientists ready to go deeper.
