Genome duplication makes snowflake yeast cells and clusters larger—but is it sufficient for the later evolution of macroscopic multicellularity?
Part of the Lab Notes research portfolio.
Interactive laboratory: https://lindgreendavid.github.io/snowflake-evolution-lab/
Engineered tetraploidy increased 24-hour cluster radius in both tested genetic backgrounds, but all evolved PA and PM populations were already approximately tetraploid while only PA became macroscopic. At transfer 1,000, mean PA radius was 8.859× mean PM radius. Whole-genome duplication helps; it is not sufficient for the later macroscopic phenotype.
The historical frozen v0.1 analysis remains unchanged: cell aspect ratio and cluster radius were positively associated within all five anaerobic lineages (Spearman ρ 0.923–0.967; median 0.956).
This project contributes a source-compatible reanalysis that separates an immediate engineered genome-duplication effect from a longitudinal sufficiency test, while preserving the earlier morphology study unchanged. Its interactive model makes permanent parent–daughter junctions, steric contact, and modelled fracture visibly distinct from measured quantities. It does not identify a quantitative entanglement threshold, recreate microscopy, or provide an independent wet-lab replication.
Replay the original transfer 0–600 morphology trajectories, then open the v1 genome explorer to compare engineered 2N/4N effects and inspect chromosome copy number, G1 peak, and cluster radius through transfer 1,000. Published summaries drive the displays. The colony animation remains an explanatory model—not microscopy or a forward evolutionary simulation—and its full mapping is documented in the current animation model specification. The historical v0.1 specification remains available unchanged.
The v1 analysis uses the public source tables for Tong et al. (2025), pinned to upstream commit
aa090bbd9163dda490a5ede716bdd063270e9cd6, with exact SHA-256 verification. Inference is performed
on four biological replicate strains per engineered group, not on thousands of segmented cells or
clusters as if they were independent. The longitudinal check follows PA1–PA5 and PM1–PM5 through
transfer 1,000. See the source audit.
python -m pip install -e '.[dev]'
snowflake-evolution
python scripts/build_site_data.py
pytestTo regenerate the compact v1 inputs from the authors' repository, clone their pinned commit and run
python scripts/import_v1_sources.py /path/to/WGD_in_MuLTEE. The importer refuses files whose
hashes do not match the frozen provenance. Generated reports and browser data must match the
committed artifacts exactly.
This is a transparent reanalysis of published MuLTEE data. The selected public tables do not contain a quantitative, time-resolved entanglement endpoint joined to each genome and morphology observation, so this project does not estimate an entanglement threshold. It is not an independent wet-lab replication, and laboratory yeast do not reproduce the historical origins of animals, plants, fungi, or other multicellular lineages.
Read the frozen v1.0 protocol, v1.0 research report, and machine-readable provenance. Historical v0.1 study artifacts remain available and unchanged.
The original analysis software and website are available under the MIT licence. Cite this software using CITATION.cff and cite the primary paper for the measurements. Upstream data retain their original attribution and terms.