Welcome to the new BRIE2 (Bayesian regression for isoform estimate, v2), a scalable Bayesian method to robustly identify splicing phenotypes in single cells RNA-seq designs and accurately estimate isoform proportions and its uncertainty.
BRIE2 supports isoform quantification for different needs:
- cell features: informative prior is learned from shared cell processes. It also allows to effectively detect splicing phenotypes by using Evidence Lower Bound gain, an approximate of Bayes factor.
- gene features: informative prior is learned from shared gene regulatory features, e.g., sequences and RNA protein binding
- no feature: use zero-mean logit-normal as uninformative prior, namely merely data deriven
Note, BRIE1 CLI is still available in this version but changed to brie1 and brie1-diff.
BRIE2 is available through PyPI. To install, type the following command
line, and add -U
for upgrading:
pip install -U brie
Alternatively, you can install from this GitHub repository for latest (often development) version by following command line
pip install -U git+https://github.com/huangyh09/brie
In either case, add --user
if you don't have the write permission for your
Python environment.
For more instructions, see the installation manual.
The full manual is at https://brie.readthedocs.io More examples and tutorials are coming soon.
In brief, you need to run brie-count first, which will return a count matrix
and hdf5 file for AnnData. Then you can use brie-quant to perform
quantification in different settings. Type command line brie-count -h
and
brie-quant -h
to see the full arguments.
- Yuanhua Huang and Guido Sanguinetti. Computational identification of splicing phenotypes from single cell transcriptomic experiments. bioRxiv, 2020; 368019.
- Yuanhua Huang and Guido Sanguinetti. BRIE: transcriptome-wide splicing quantification in single cells. Genome Biology, 2017; 18(1):123.