First Study Demonstrates Scaling Laws for Transformer Models in Single-Cell Genomics
Researchers have published the first systematic study demonstrating that neural scaling laws — power-law relationships between model size, data, and loss — emerge in masked-reconstruction transformers trained on single-cell RNA sequencing data. The study tested seven model sizes spanning three orders of magnitude in parameter count using data from the CELLxGENE Census, finding clear scaling behavior only when sufficient data was available. The findings have practical implications for designing large-scale 'foundation models' for single-cell biology.
A preprint posted to arXiv presents the first systematic investigation of whether neural scaling laws, well-established in language and vision AI, also hold for transformers applied to single-cell RNA sequencing (scRNA-seq) data. The researchers constructed two experimental regimes: a data-rich setting (512 highly variable genes, 200,000 cells) and a data-limited setting (1,024 genes, 10,000 cells), then trained models ranging from roughly 533 to 340 million parameters. In the data-rich regime, validation mean squared error followed a clear power-law relationship with model size, with an irreducible loss floor estimated at approximately 1.44 MSE, or about 2.30 bits of entropy per masked gene position in information-theoretic terms. In the data-limited regime, scaling was negligible, indicating that lack of data — not model capacity — is the binding constraint in low-data settings. The authors conclude that the data-to-parameter ratio is a critical factor governing whether scaling benefits materialize, and they outline further measurements needed to refine the entropy estimate and guide the development of single-cell genomics foundation models.
What's missing
The study is a preprint and has not yet undergone formal peer review. The authors acknowledge that the entropy estimate of ~2.30 bits per masked gene position is preliminary and requires additional measurements to validate. The study does not assess downstream biological task performance, so it remains unclear whether scaling in reconstruction loss translates to improved performance on practical genomics applications. The experiments are limited to two specific data regimes and one architecture type (masked-reconstruction transformers), leaving open questions about generalizability to other model families or data modalities.
What different sources said
- arXiv cs.LGCenter
Scaling Laws for Masked-Reconstruction Transformers on Single-Cell Transcriptomics
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.