Novel Methods for Identification of Multi-Dimensional Co-Exclusion Patterns in Oral Microbiomes

ABSTRACTOral health plays an important role in overall human health and well-being. The mouth is the mostmicrobiologically diverse environment in the body with 700?800 bacterial species identified from the humanoral microbiome to date. As such, the role of human mouth (gum, teeth) associated microbial communities inoverall oral health cannot be underestimated. Understanding inter-species relationships among members ofmicrobial communities is key to understand and control the microbiota. Fortunately, significant improvements inHigh Throughput Sequencing (HTS) technology have led to a dramatic increase in the number of studiesfocused on microbiomes. This, in turn, has led to ongoing efforts to develop new tools and methods for theanalysis of HTS based microbiome data in the hope to advance from simple identification of microorganismsunder/over represented in specific groups of samples to detection of complex patterns between abundances ofdifferent microorganisms and understanding relationships among members of microbial communities (e.g.,commensalism, mutualism and amensalism). Many such efforts have been focused on using correlation tocharacterize the strength of pairwise co-occurrence patterns. Co-exclusion is arguably one of the mostimportant patterns reflecting microorganisms' intolerance to each other's presence. Knowing these relationsopens an opportunity to manipulate microbiota, personalize anti-microbial and probiotic treatments as well asopen the possibility of microbiota transplantation in the future. The co-exclusion patterns, however, cannot beappropriately identified by existing methods. The overall goal of the proposed research is to develop a novelway to identify and evaluate the statistical significance of co-exclusion patterns between two, three or morevariables describing a microbiota and allow one to extend analysis beyond microorganism abundance byincluding other microbiome associated measurements such as, pH, temperature, etc., as well as estimate theexpected numbers of false positive co-exclusion patterns in a co-exclusion network.