Using WEKA to Classify Treponemes Based on Carbohydrate Utilization and Enzymatic Activity

Abstract

Prediction and classification of microbial species is an important skill for any clinical, industrial or environmental laboratory, there are many aspects including carbohydrate utilization, enzymatic activity that can be used for predicting with high accuracy bacterial species. The unique features presented by machine learning includes the possibility of classify and predict bacterial species based on their biochemical or enzymatic activities though decision tree classification, data visualization, clustering and neural network among others tools. In this research, decision trees are applied to classify Treponema species based on enzymatic activity, visualization tools showed comparisons among multiple biochemical aspects and neural networks created patterns for carbohydrate utilization. Treponema species are invasive pathogens causing a range of significant clinical pathologies in many cases ending in neurological complications such as in the case of syphilis produced by Treponema pallidum. Previous research papers explored the used of PyBact to generate a matrix which it is then evaluated through machine learning resulting in a high percentage of correct classification. Our findings indicate that decision trees are one of the most effective tools to classify bacterial species contributing significantly to any medical or diagnostic laboratory. There are several ML applications that remain to be explored not just for a particular genus but with questions involving the human microbiome, biofilm formations, and the current COVID 19 pandemic.

Keywords: Machine learning, PyBact, neural networks, decision trees, visualization, Treponema genus.

DOI: 10.7176/JNSR/13-6-04

Publication date:March 31^st 2022