Genomic epidemiology analysis: Serial interval distribution and cluster assignment

Genomic epidemiology has gained significant popularity since the global COVID-19 outbreak. It combines the power of advanced genomic sequencing and epidemiology to gain valuable insights into the transmission dynamics during epidemics. In the context of infectious disease outbreaks, genomic epidemiology plays a crucial role. By sequencing the genomes of pathogens that are responsible for infectious diseases, healthcare professionals can gain valuable insights into the transmission and evolution of these pathogens, which is useful in identifying the specific strain of a pathogen causing the outbreak, tracking its spread and evolution over time, and guiding the response with public health interventions such as vaccination campaigns or quarantine measures. In this thesis, we will highlight the importance of genomic epidemiology in estimating the serial interval distribution and predicting the genomic clustering (as these are important keys in characterizing an outbreak) when we have incomplete sampling data. In the first half, we present a novel approach to estimate the serial interval distribution when the information of who infected whom is not available. Here, we will show how to utilize viral sequence data as an integral to rule out the potential infectors of a case. We assess our method's performance on simulated data sets as well as several real-life outbreaks, and implement our approach on the 2020 COVID-19 outbreak in Victoria, Australia. In the second half, we develop a method to assign unsequenced cases (infected individuals) to clusters assuming that a more direct method of linking individuals, such as contact tracing, is not available. We consider a problem of cluster assignment where genomic data is incomplete for a subset of cases due to, for example, cost, sample quality, or other constraints. We will show how to leverage additional sources of data, such as demographic, epidemiological, and clinical data, in order to assign unsequenced cases to clusters as well as to address some questions related to clustering. We demonstrate our method on the 2014-2016 Tuberculosis outbreak in Valencia, Spain.