Skip to main content
Spatial proteomics from H&E. Lattice predicts spatially resolved protein-marker maps from routine H&E, delivered as virtual mIF channels. Each channel is a per-pixel model output for one protein marker. Experimental multiplex measurements remain the training and comparison reference. Lattice is trained on paired H&E and multiplex immunofluorescence data. The platform accepts a slide, runs inference on our infrastructure, and returns predicted marker channels you can use with the rest of your spatial pipeline.

​
Why Strand AI

H&E in, markers out

You don’t need antibody panels, IF capacity, or a second slide. Submit the H&E and request the markers you need.

Spatial outputs

Predictions are returned as multi-channel OME-Zarr you can open as AnnData (Python) or SpatialExperiment (R) and work with using your existing multiplex tooling.

REST + SDKs

Python and R clients, a documented REST API, and a credits ledger you can estimate against before you submit.

​
Who this is for

Researchers and biotech teams who have H&E at scale and want predicted spatial proteomics, typically for:
  • biomarker hypothesis generation in retrospective cohorts,
  • enriching slides where IF or mIF was not collected,
  • batch-level QC of multiplex panels against an orthogonal predictor.
Lattice is trained primarily on oncology data. Predictions outside an oncology context are exploratory; if you have a non-oncology use case, email us.

​
What’s next

Platform quickstart

Sign in, upload an H&E slide, run Lattice, and inspect the marker layers.

SDKs

Use Python, R, or the strand command line for scripted workflows.

MCP server

Connect Claude or another MCP client to a Strand organization.

API reference

Build directly against the canonical HTTP surface.
Strand AI is in beta. You can sign up at app.strandai.com and run the public marker panel on your own slides. Email support@strandai.com to discuss an expanded panel or a larger cohort.
For research use only. Strand AI predictions are model outputs intended for research and hypothesis generation. They are not validated for, and must not be used in, clinical diagnosis, treatment selection, or patient care decisions.