A0955
Title: A branching process model for digital read quantification
Authors: Karin Dorman - Iowa State University (United States) [presenting]
Abstract: Sequenced read counts are a ubiquitous data summary of modern high-throughput biological methods used to observe metagenomes, genomes, transcriptomes, epigenomes, and various kinds of molecular interactions and functions. Almost all such count data are obtained after amplification of sampled molecules, which can bias and overdisperse the biological signal of interest. The purpose is to develop and investigate a novel model for count data that better adheres to the experimental generative process than Poisson and negative binomial models with or without zero-inflation. The model is based on a branching process model of polymerase chain reaction (PCR) amplification. It naturally accounts for overdispersion and zero inflation, with meaningful parameters directly linked to biological processes. In particular, the first estimates of PCR amplification efficiency are provided during library preparation and estimate the effects of primer mismatch on sampling efficiency.
