MOTIVATION: A growing body of literature has demonstrated the potential for non-invasive diagnosis of a variety of human genetic diseases using cell-free DNA extracted from maternal plasma samples in early gestation. Such methods are of great significance to the obstetrics community because of their potential use as clinical standard of care. Proof of concept for such approaches has been established for aneuploidy and paternally inherited dominant traits. Although significant progress has recently been made, the non-invasive diagnosis of monogenic diseases that segregate in a recessive mendelian fashion is more problematic. Recent developments in microfluidic digital PCR and DNA sequencing have resulted in a number of recent advances in this field. These have largely, although not exclusively, been used for the development of diagnostic methods for aneuploidy. However, given their prevalence, it is likely that such methods will be utilized towards the development of non-invasive methods for diagnosing monogenetic disorders. RESULTS: With this in mind, we have undertaken a statistical modeling of three contemporary (digital) analytical methods in the context of prenatal diagnosis using cell free DNA for monogenic diseases that segregate in a recessive mendelian fashion. We provide an experimental framework for the future development of diagnostic methods in this context that should be considered when designing molecular assays that seek to establish proof of concept in this field.

There is currently great interest in the development of methods for the minimally invasive diagnosis of fetal genetic disease using cell-free DNA from maternal plasma samples obtained in the first trimester of pregnancy. With the rapid development of high-throughput sequencing technology, the possibility of detecting the presence of trisomy fetal genomes in the maternal plasma DNA sample has recently been explored. The major concern of this whole genome sequencing approach is that, while detecting the karyotype of the fetal genome from the maternal plasma requires extremely high accuracy of copy number estimation, the majority of the available high-throughput sequencing technologies require polymerase chain reaction (PCR) and are subject to the substantial bias that is inherent to the PCR process. We introduce a novel and sophisticated statistical model for the whole genome sequencing data, and based on this model, develop a highly sensitive method of Minimally Invasive Karyotyping (MINK) for the diagnosis of the fetal genetic disease. Specifically we demonstrate, by applying our statistical method to ultra high-throughput whole sequencing data, that trisomy 21 can be detected in a minor ('fetal') genome when it is mixed into a major ('maternal') background genome at frequencies as low as 5%. This observation provides additional proof of concept and justification for the further development of this method towards its eventual clinical application. Here, we describe the statistical and experimental methods that illustrate this approach and discuss future directions for technical development and potential clinical applications.