baymobil: A Python package for detection of graft-mobile mRNA using exact Bayesian inference on RNA-Seq data

Background: A popular method for the detection of graft-mobile mRNA on a genomic scale in plants is to perform RNA-Seq on heterografts comprising different ecotypes, species, or cultivars (types). Transcripts from one plant type that are detected in the other type can be assumed to have been transported across the graft junction. A necessary step is the ability to differentiate between transcripts from each plant type, which can be achieved based on known single nucleotide polymorphisms (SNPs) between types. One current challenge is to differentiate between RNA-Seq reads associated with specific SNPs and RNA-Seq errors. Here, we present baymobil, a Python package that implements a Bayesian framework for determining between graft-mobile transcripts and sequencing errors. Results: : baymobil takes processed RNA-Seq data from homo- and heterografted plant samples as input and for each mRNA calculates a Bayes factor: the ratio of the evidence for transcripts having crossed the graft junction over the evidence for the data being a consequence of sequencing errors. These Bayes factors can be used to rank mRNAs based on the statistical support for their mobility. Furthermore, the package includes functions for the creation of simulated RNA-Seq datasets, which have been used to demonstrate that this approach outperforms existing mobility criteria and is a necessary step in the successful detection of graft-mobile mRNA. Conclusion: The baymobil Python package improves the accuracy of pipelines for the detection of graft-mobile mRNA based on SNP differences between heterografted types. It is openly and freely available via GitHub, and is easily installed with pip and pypi. A detailed tutorial with test data and results for the statistics and simulation package baymobil can be found on github.com/mtomtom/baymobil.