Postman: predicting spatial protein expression from pathology slides

Hey HN, we're Oded and Yue from Strand AI (YC W26).

H&E is the most common tissue assay in medicine. Every biopsy gets this stain. But it only shows you cell morphology: nuclei and cytoplasm. To see which proteins are expressed and where in the tissue, you need spatial proteomics (CODEX, multiplex immunofluorescence), which costs thousands per slide, takes days, and destroys the tissue.

The key insight is that lots of spatial protein signal is already encoded in the H&E. Cell shape, microenvironment architecture, patterns of immune infiltration all reflect the underlying molecular state at a level far beyond what even the best trained eyes, pathologists, can resolve.

Today we're introducing POSTMAN. It predicts spatially-resolved protein expression directly from routine H&E slides. What used to require thousands of dollars and days of specialized lab work, you can now get from the slides already sitting in your archive.

That means you can go back to existing patient cohorts, surface protein signals you never measured, and characterize the tumor microenvironment from routine pathology at a fraction of the cost.

We're opening early access through a design partner program for pharma, biotech, and research teams.

Happy to answer questions about the approach or validation.