The tracking of lineage frequencies via DNA barcode sequencing enables the quantification of microbial fitness. However, experimental noise coming from biotic and abiotic sources complicates the computation of a reliable inference. We present a Bayesian pipeline to infer relative microbial fitness from high-throughput lineage tracking assays. Our model accounts for multiple sources of noise and propagates uncertainties throughout all parameters in a systematic way. Furthermore, using modern variational inference methods based on automatic differentiation, we are able to scale the inference to a large number of unique barcodes. We extend this core model to analyze multi-environment assays, replicate experiments, and barcodes linked to genotypes. On simulations, our method recovers known parameters within posterior credible intervals. This work provides a generalizable Bayesian framework to analyze lineage tracking experiments. The accompanying open-source software library enables the adoption of principled statistical methods in experimental evolution.

Over 800 million people around the world share their social interactions with friends on Facebook, providing a rich body of information referred to as the social graph. In this talk, I describe how we model and serve this graph. Our model uses typed nodes (fbobjects) and edges (associations) to express the relationships and actions that happen on Facebook. We access the graph via a simple API that provides queries over the set of same-typed associations leaving an object. We have found this API to be both sufficiently expressive and amenable to a scalable implementation. In the last segment of the talk I describe the design of TAO, our graph data store. TAO is a distributed implementation of the fbobject and association API that has been serving production traffic at Facebook for more than 2 years.