Automated Ribosomal Intergenic Spacer Analysis (ARISA) is one of the community profiling techniques to study microbial community composition and its variation in environmental ecology. ARISA uses 16S-23S rRNA intergenic spacer length heterogeneity at different times and places. As large amount of community profiling data are generated by these approaches, effective analytical approaches for the profiles are essential to draw meaningful conclusions. This dissertation focuses on computational approaches for bacterial community profiling data analysis and it mainly contains two parts:

Part I. Data preprocessing. Due to errors from various sources, the profiling data read directly from the fragment analyzer need to be properly binned before any down stream statistical analysis. We developed a dynamic programming algorithm (DPA) based binning approach for binning ARISA profiling data which minimizes the overall differences between replicates from the same sampling location and time. Clustering analysis of the ARISA from different times based on the dynamic programming algorithm binned data revealed important features of the biodiversity of the microbial communities. The robustness of DPA binning, i.e. how stable the bins are as more data come in, was also discussed.

Part II. Interaction analysis. In marine ecosystems, interactions between species are complex and dynamic. These interactions are often missed by Pearson Correlation Coefficient (PCC) analysis because PCC is good at capturing the linear relationships between two species. We introduced two approaches: Local Similarity Analysis (LSA) and Liquid Association (LA), to identify more complex dependent associations among species as well as associations between species and environmental factors. The results, combined with results from PCC analysis were used to construct a theoretical ecological network which allows for easy visualization of the most significant associations.

In summary, several computational approaches are introduced to analyze microbial community profiles from culture independent techniques for better characterization of the natural community composition and better understanding of the ecological dynamics in the microbial community.