Microorganism species can become pathogenic and cause bacterial infection, result from the imbalance in microbial ecosystem between host and microbe. Microorganisms emit secondary metabolites, known as volatile metabolites or organic compounds (VOCs) for various functions such as intra- or inter-species interactions, defense and attraction. Currently, VOCs are widely used as a biomarker for human diseases. This research is aimed to identify the relationship between microorganism species and volatile metabolite compound from the collected species and VOCs emitted organism data by using unsupervised machine learning approaches such as hierarchical clustering and graph-clustering method. Supervised machine learning methods also been used to classify the microorganism pathogenicity such as support vector machine (SVM) and random forest (RF). These data are collected from KNApSAcK and mVOC database where it provides most of the microorganism species and metabolites contents. From the collected data, there are in total of 1088 VOCs emitted by 517 microorganism species. As a result, the application of machine learning methods enable us to identify the relationship of species with their emitted VOCs and classify the microorganism species into their own pathogenicity.
展开▼
