Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.strandai.com/llms.txt

Use this file to discover all available pages before exploring further.

Strand AI predicts protein markers directly from an H&E whole-slide image using POSTMAN, our spatial-proteomics imputation model. Each marker listed below has a model checkpoint trained against paired multiplex ground truth. To request a marker not on this list, email support@strandai.com. We add markers as ground-truth training data becomes available.
Per-marker model cards with calibration plots, per-tissue accuracy, and held-out benchmarks are coming soon. The table below is a snapshot of what’s predictable today.

Marker panel

All 52 markers below are served by postman-v1. Grouped by biology; expand any group to see the full list.

Immune lineage (18)

MarkerPredicts
CD3ePan-T cell marker.
CD4Helper T cells.
CD8Cytotoxic T cells.
CD45Pan-leukocyte.
CD45RANaive T cells (PTPRC isoform).
CD45ROMemory T cells (PTPRC isoform).
CD20Mature B cells.
CD21Mature B cells and follicular dendritic cells.
CD79B cell receptor signaling.
CD11bMyeloid / neutrophils / monocytes.
CD11cDendritic cell / myeloid lineage marker.
CD14Monocytes and tissue macrophages.
CD68Pan-macrophage.
CD163M2-polarised macrophages.
CD141Endothelial and BDCA-3 dendritic cell subset.
CD66Granulocytes and select epithelial cells (CEACAM family).
MPONeutrophil granule enzyme.
FoxP3Regulatory T cells.
MarkerPredicts
PD1T-cell exhaustion / checkpoint.
PDL1Tumor and immune PD-L1 expression.
LAG3T cell exhaustion / checkpoint.
VISTAT cell suppression / checkpoint.
ICOST cell costimulation.
CD38B, T, and plasma cell activation.
CD39T-regulatory / endothelial ectonucleotidase.
CD40B cell activation and antigen-presenting cells.
CD44Hyaluronan receptor; broad immune and tumor expression.
IDO1Immunosuppressive enzyme.
GranzymeBCytotoxic effector function.
IFNgInterferon-γ; activated T and NK cells.
HLA-DRMHC class II; antigen-presenting cells.
HLA-ABCClassical MHC class I.
HLA-ENon-classical MHC class I; NK cell inhibition.
MarkerPredicts
DAPINuclear counterstain (sanity-check channel).
Ki67Proliferation marker.
PCNAProliferation marker (DNA replication).
MarkerPredicts
PanCKPan-cytokeratin; epithelial / tumor compartment.
EpCAMPan-epithelial adhesion marker.
ECadE-cadherin; epithelial cell-cell adhesion.
Keratin8/18Simple epithelial cytokeratins.
TP63Squamous and basal epithelial cells.
GATA3TH2 / luminal epithelial transcription factor.
MarkerPredicts
aSMAα-smooth muscle actin; stromal myofibroblasts and vasculature.
VimentinMesenchymal marker; stromal and EMT-state cells.
Caveolin1Caveolae scaffold; stromal and tumor compartments.
CD31PECAM-1; endothelium and microvasculature.
CD34Endothelial and hematopoietic stem / progenitor cells.
CollagenIVBasement membrane.
PodoplaninLymphatic endothelium and cancer-associated stroma.
MarkerPredicts
BCL2Apoptosis regulator.
Gal3Galectin-3; immune and stromal modulation.
PGP9.5Pan-neuronal / nerve fibers.

Reading the predictions

Predictions are returned as a multi-channel OME-Zarr aligned to the slide’s pixel grid. Each requested marker becomes a separate channel; the SDK helpers convert it to AnnData (Python) or SpatialExperiment (R) so you can treat it like real multiplex data. See the Quickstart for an end-to-end example and the Python / R SDK reference for the conversion helpers.

Accuracy caveats

  • Predictions are model outputs, not ground truth. Use them as a hypothesis-generation surface, particularly on cohorts that look out-of-distribution to the training data.
  • Per-marker confidence varies. Lineage and structural markers (e.g. PanCK, CD31, aSMA) generally calibrate better than functional / activation markers (e.g. GranzymeB, PD1).
  • We do not yet publish per-tissue or per-organ accuracy numbers. Those ship with the per-marker model cards.
Reach out at support@strandai.com if you want pre-publication benchmarks on a specific tissue or marker.