|
Abstract:
|
At least two of the working groups in the SAMSI Beyond Bioinformatics program examined research problems arising in the analysis of human microbiome data. We concentrate on methods used in the analysis of vaginal microbiome data from the Virginia Commonwealth University's NIH-HMP project, with the goal of understanding the clinical condition of bacterial vaginosis (BV) as it correlates with the microbiome composition. A common first step is to filter and normalize microbiome data, and to use ordination methods to obtain lower dimensional representations of the data for visualizing patterns in vaginal microbiome samples with respect to clinical conditions. The effects of filtering and normalizing on ordination methods were explored in the SAMSI working groups. In addition, examples and tools from different programs and pipelines that allow for improved visualization were developed and consolidated for use with human microbiome data. In this talk, we present work on the effects of normalization and filtering on ordination methods using data from the VCU studies, while also demonstrating use of the tools for visualization.
|
