"Why wont this run again?!": Making R analysis more reproducible