“Genetic data is always imperfect. There are missing genotypes, wrongly labeled bits. You can mitigate some of those problems [with different algorithms that fill in the blanks] but if you change the method you use to mitigate it, you need to know exactly why it changed.”

"Pachyderm is a seamless complement to our pipeline and processes. It already has built-in, ready-to-go systems for managing commit history, automatically tracking datum status, and it can reliably restart a process and keep the state of the overall pipeline consistent. And, of course, everyone has experienced an inability to roll …