Microbial organisms are abundant in nature. Even in the human body the amount of microbes present is much higher than the number of individual cells the human body is composed of. Hence a clear knowledge about microbes and their nature is exceptionally important. In this Thesis Proposal we explore the computational algorithms and optimisation techniques which can be applied to recognise patterns in longitudinal data in the domain of microbiology, specifically on analysing and inferring how microbes interact with each other. We work mainly with microbial abundance profiles which are generated by analysing the 16S ribosomal RNA in in-vitro samples to identify the taxonomic information, across multiple time points.
First, we successfully use a Genetic Algorithm based method, coupled with a Microbial Community Dynamics model to infer the interactions in microbial communities in human body sites. Next we propose that microbial interactions could be time varying and they possibly vary according to a pattern. We show preliminary results which affirm our assumptions. Lastly, we propose that parallel data-sets can be used to enhance the lack of data which is generally an issue in recognising patterns in microbial data, especially for studies of shorter duration.