A bit-packer for DNA methylation data — arrays and whole genomes, where analysis is bitwise, and so stays fast from 28K probes to 29M CpGs.
For detailed documentation, tutorials, and usage examples, visit the YAME User Guide.
YAME packs DNA methylation into bits. A family of compact binary formats (CX formats) holds methylation values, MU counts, categorical states, fractions, masks and coordinate streams — as little as one bit per CpG — inside a single record layout.
The payoff is that questions become bit operations. A data file stores no coordinates: row i means whatever row i of the reference means, so intersecting a methylome with a feature set is a bitwise AND rather than a genomic join. An Infinium manifest and a whole-genome CpG set are the same kind of object under that rule — a row space, differing only in length — which is why the same commands serve array and sequencing data.
- Bit-level packing: 1 bit per CpG for binary calls, 2 for set/universe — a whole-genome hg38 track is 3.5 MB
- Array and sequencing alike: hg38, mm10, mm39, MSA, EPICv2, EPIC, HM450, HM27, MM285, Mammal40
- Scalable to hundreds of thousands of single cells
- Versatile data support: MU counts, binary methylation, chromatin states, fractions, differential calls, and CpG coordinate streams
- Comprehensive toolkit: packing, unpacking, downsampling, subsetting, row operations, enrichment testing, and summarization
- Consistent internal API: all data stored as
cdata_tblocks inside BGZF frames - Integrates seamlessly with bedtools, KYCGKB, and other methylation workflows
conda install -c zhou-lab -c conda-forge yameThe zhou-lab channel is published by CI on every release tag, so it is
always current. The bioconda recipe
lags well behind and its build predates yame fetch, so it cannot download
the reference data the other commands resolve -R and -m against — prefer
the channel above until that catches up.
If you use YAME in your research, please cite:
Goldberg*, Fu*, Atkins, Moyer, Lee, Deng, Zhou† (2025). "KnowYourCG: Facilitating Base-level Sparse Methylome Interpretation." Science Advances. https://doi.org/10.1126/sciadv.adw3027
- Documentation: https://zhou-lab.github.io/YAME/
- Issues: Please report bugs and feature requests on the GitHub Issues page
YAME is dual-licensed:
- AGPL-3.0 for academic, educational, and non-profit research use
- Commercial License for commercial applications
YAME is free to use for academic research, educational purposes, and non-profit organizations under the GNU Affero General Public License v3.0 (AGPL-3.0).
If you wish to use YAME in commercial products or services, or if the AGPL-3.0 restrictions are not suitable for your use case, please contact us for a commercial license: [zhouw3@chop.edu]
Developed by the Zhou Lab