bakR - Analyze and Compare Nucleotide Recoding RNA Sequencing Datasets
Several implementations of a novel Bayesian hierarchical
statistical model of nucleotide recoding RNA-seq experiments
(NR-seq; TimeLapse-seq, SLAM-seq, TUC-seq, etc.) for analyzing
and comparing NR-seq datasets (see 'Vock and Simon' (2023)
<doi:10.1261/rna.079451.122>). NR-seq is a powerful extension
of RNA-seq that provides information about the kinetics of RNA
metabolism (e.g., RNA degradation rate constants), which is
notably lacking in standard RNA-seq data. The statistical model
makes maximal use of these high-throughput datasets by sharing
information across transcripts to significantly improve
uncertainty quantification and increase statistical power.
'bakR' includes a maximally efficient implementation of this
model for conservative initial investigations of datasets.
'bakR' also provides more highly powered implementations using
the probabilistic programming language 'Stan' to sample from
the full posterior distribution. 'bakR' performs multiple-test
adjusted statistical inference with the output of these model
implementations to help biologists separate signal from
background. Methods to automatically visualize key results and
detect batch effects are also provided.