Ultrasensitive Detection of Rare Variants in Admixture Populations Using Next-Generation Sequencing

• Activity Number: 480
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Patrick Flaherty*+ and Georges Natsoulis and Omkar Muralidharan and Nancy Zhang and Jason Buenrostro and John Bell and Mark Winters and Mark Holodniy and Hanlee Ji
• Companies: Stanford University and Stanford University and Stanford University and Stanford University and Stanford University and Stanford University and Stanford University and VA Palo Alto Healthcare System and Stanford University
• Keywords: next-generation sequencing ; genomics ; viral ; hierarchical model

Abstract

The viral load of an infected individual often contains a population of quasi-species each with a different genome. A rare resistant variant may survive lethal selection and proliferate causing prolonged illness. We describe improvements to both statistical and experimental methodology using next-generation sequencing to identify rare mutational events. We develop a hierarchical model to account for over-dispersion in the observed data. We show a theoretical power analysis involving detection limit improvements due to both sample preparation precision and read depth. We present results on the detection of a quasi-species present at a prevalence of 0.1% with high sensitivity (100%) and specificity (99%) using an experimentally defined admixed population of synthesized DNA.