Adi L. Tarca, PhD
Professor of Obstetrics and Gynecology and of Computer Science, Section Head of the Bioinformatics and Computational Biology Unit, Perinatal Research Initiative in Support of the Perinatology Research Branch

Abstract

Development of treatment and prevention strategies for obstetrical syndromes, such as spontaneous preterm birth and preeclampsia, require early identification of women at risk. Ideally, predictive strategies should also pinpoint possible mechanism of disease leading to the phenotypic manifestation. Genomics based approaches have been proposed for these goals, yet molecular signatures often fail to generalize to new cohorts and have low accuracy. Persistent challenges include the lack of reliable molecular readouts in non-invasive samples to inform of patient specific risk, and multiple etiologies of disease contributing heterogeneity in the population. These issues are compounded by the typically low number of samples available relative to the number of omics features being measured. It this talk, I will illustrate the use of transcriptomics and proteomics in maternal blood for prediction of obstetrical disease, including results from the DREAM Preterm Birth Prediction Challenge. This crowdsourcing initiative that we have organized, assessed the accuracy of longitudinal molecular profiles for predicting spontaneous preterm birth in asymptomatic patients, and identified optimal analytical pipelines. Strategies to address the current challenges in the field and develop more robust disease signatures will also be discussed. These include integration of evidence from different bio-sources, and disaggregation of disease groups by indicators of pathophysiology.