simpleCache: Simply Caching R Objects

title: 'simpleCache: R caching for reproducible, distributed, large-scale projects' authors: - name: Nathan C. Sheffield orcid: 0000-0001-5643-4068 affiliation: 1 - name: VP Nagraj orcid: 0000-0003-0060-566X affiliation: 1 - name: Vince Reuter orcid: 0000-0002-7967-976X affiliation: 1 affiliations: - name: University of Virginia index: 1 date: 28 October 2017 bibliography: paper.bib

Summary

simpleCache is an R[@R] package that provides functions for caching R objects. Its purpose is to encourage writing reusable, restartable, and reproducible analysis for projects with large data and computational requirements. Like its name indicates, simpleCache is intended to be simple. Users specify a location to store caches, and then provide nothing more than a cache name and instructions (R code) for how to produce an R object. simpleCache either creates and saves or simply loads the result as necessary with just a single function call.

In addition to this basic functionality, simpleCache has advanced options for assigning objects to specific environments, recreating caches, reloading caches, and even distributing caching operations to cluster computing resources via the batchools[@batchtools] interface. These features make the package particularly useful for large-scale data analysis and research projects. simpleCache is most helpful for caching objects that are computationally expensive to create, but used in multiple scripts or by multiple users.

simpleCache is also useful to enhance performance in a package that relies on large databases. For example, simpleCache has been incorporated with the LOLA R package[@LOLA] to more efficiently cache and retrieve genomic region databases. Similarly, simpleCache has been used to store cached baseline statistical tables for faster lookup to determine statistical differences on tables with hundreds of millions of data points [@RPIM].

In summary, simpleCache provides a user-friendly interface to help the R programmer manage computationally intensive, repeated data analysis.