atlasqtl is an R package implementing a scalable hierarchical modelling framework for variable selection in regression problems with many predictors and many responses. The method is tailored to the detection of hotspots, i.e., predictors associated with several responses. The hotspot propensity of each candidate predictor is modelled using a horseshoe distribution (Carvalho et al. 2009), whose local scale flexibly models the large hotspot effects and whose global scale adapts to the overall signal sparsity. Inference is performed using efficient batch variational inference updates which are coupled with simulated annealing schemes to improve the exploration of multimodal parameter spaces.
The method can be employed in any sparse multiple-response regression setting, and is particularly suited to large molecular quantitative trait locus (QTL) problems, in which hotspot genetic variants, controlling many molecular levels at once, may be responsible for decisive regulatory mechanisms. Hence, our approach is a tool that can help towards understanding the functional architecture underlying complex traits and outlining a hotspot atlas of the human genome.
Reference: H. Ruffieux, A. C. Davison, J. Hager, J. Inshaw, B. Fairfax, S. Richardson, and L. Bottolo. A global-local approach for detecting hotspots in multiple response regression. The Annals of Applied Statistics, 14:905-928, 2020.
Important note: the R package depends on the GSL library which needs to be manually installed. For example on Ubuntu,
$ sudo apt-get install libgsl-devor on mac,
$ brew install gslafter having installed Homebrew.
Then, to install the package in R, run the following command:
if(!require(remotes)) install.packages("remotes")
remotes::install_github("hruffieux/atlasqtl")This software uses the GPL v3 license, see LICENSE. Authors and copyright are provided in DESCRIPTION.
To report an issue, please use the atlasqtl issue tracker at github.com.