pychiral is a Python implementation of CHIRAL (Circular HIerarchical Reconstruction ALgorithm), an R package originally developed by Lorenzo Talamanca, designed to infer circadian clock phases from RNA-seq data. This algorithm is based on Expectation Maximization and utilizes a Statistical Physics-inspired approach.
- posteriors via grid eval
- PCA init
- Efficiently infers circadian clock phases from RNA-seq data.
- Incorporates Expectation Maximization algorithm.
- Inspired by methods from Statistical Physics.
This method was applied in the following Science paper: Sex-dimorphic and age-dependent organization of 24-hour gene expression rhythms in humans. Further information regarding the alorithm can be found in the supplement.
For the original R version of this package, visit the CHIRAL repository here.
To run pychiral, you need to install the following dependencies:
pip install .pychiral is compatible with anndata, and it receives data in the format Nsamples x Ngenes (unlike the Bulk conventions). A layer of the dataset needs to be specified by argument. The layer passed to CHIRAL needs to be log-tranformed. By default the data will also be mean centered and standardized.