Most new spatial transcriptomics projects reach for a single-cell foundation model. They can hit the same three walls trying to integrate it.
The Data Preprocessing Trap
The model is not the only bottleneck. Data preprocessing can consume much of the effort for scGPT-family models, and “normalize the counts” hides consequential analytical choices.
Preprocessing for scGPT or SATURN, in practice: raw count matrices from 10x Genomics, SCTransform for normalization, drop cells with fewer than about 200 detected genes and cells whose mitochondrial-read fraction runs too high (commonly above 5 to 20 percent), doublet detection with DoubletFinder, Harmony for batch correction, PCA, reduce to the top 2,000 highly variable genes. Each step has parameters that matter. Each one introduces artifacts if you get them wrong.
Most teams treat this as a pipeline problem. It is not. It is a biological-interpretation problem masked as a data-science problem. The mitochondrial filtering threshold depends on whether you are working with immune cells (high mitochondrial activity) or fibroblasts (low). The batch-correction strategy depends on whether your batches represent technical replicates or biological conditions. Get any of these wrong and you are feeding garbage into a sophisticated foundation model.
Published notebooks cannot capture every dataset-specific decision or failed parameter choice. Applying the same code to different biological material still requires validating that filtering and correction have not removed the signal under study.
The Hardware Reality Check
The published training-corpus scales are real. A typical biotech R&D team does not have eight A100 80GB GPUs sitting around. The decision point is not whether to use a foundation model. It is whether to fine-tune (cheap but limited) or run inference-only (free but inflexible).
Published numbers from large single-cell foundation-model groups put training-from-scratch in the neighborhood of 8× A100 80GB for ~72 hours, and fine-tuning on your own spatial transcriptomics data in the neighborhood of 4× A100s for ~12 hours. Most biotechs do not have that infrastructure. They have a couple of V100s in the cloud that they use for standard analysis pipelines.
You have two choices.
Fine-tune a smaller version of the model. This works if your domain is close to the training data: mouse brain tissue, human tumor microenvironments. It fails catastrophically if you are working with something exotic like engineered microbial communities or plant tissue. The model learns on mouse and you are asking it to understand algae. The gap is too wide.
Use the published checkpoint as inference-only. This is free and accessible from a single V100, but you are stuck with what is encoded in the model. If the checkpoint does not know about your specific cell type or treatment condition, those embeddings will be noisy. You can get basic cell-type annotation out of it, but it will not discover novel phenotypes that were not in its training corpus.
Most teams start with inference-only, realize it is too limiting for their actual research question, then scramble to justify the GPU budget. That scrambling happens three months into a six-month project timeline.
The Regulatory Wall
The “foundation model” framing borrowed from NLP does not map cleanly to regulated validation. In NLP, the cost of a wrong word is low and we accept that foundation models will sometimes be wrong. In diagnostics, the cost of a wrong cell classification can be high enough to stop a program.
For regulated applications, a predicted subpopulation may still need validation with orthogonal methods such as flow cytometry, immunohistochemistry, or single-cell proteomics. A foundation model can accelerate discovery; it does not eliminate the validation burden.
Define what “fit for purpose” means at the start, not at submission.
The Foundation Model Is Not the Moat
The right question is not “should we use a single-cell foundation model.” It is “do we have a single-cell use case that is bigger than what a well-trained PhD with scanpy can answer in two weeks?”
Often the answer is no. Differential expression, cell-type annotation, trajectory inference: these are still faster and more interpretable with traditional methods. Foundation models earn their cost when you have massive multi-dataset meta-analysis problems, or when you are integrating modalities that do not normally play well together (spatial + single-cell + proteomics).
The threshold is roughly this: if you can solve your biological question with fewer than 50,000 cells from a single experiment, use traditional methods. If you need to integrate across five or more datasets, multiple modalities, or thousands of conditions, a foundation model starts to make sense.
What separates the teams that get value from these models is not biological sophistication. It is operational discipline. Dedicated data engineers who understand both the biology and the computational requirements. Preprocessing pipelines tuned to specific cell types and conditions. GPU resources allocated before the project starts, not as an afterthought. A clear position on which regulatory regime the output will eventually face.
The foundation model is a component. The pipeline that makes it reproducible, interpretable, and fit for purpose is what delivers value. Start with the pipeline and add the model only when the scale justifies it.
That sequence, pipeline first and model second, informs the approach at McIntosh Consulting.