NetworkParser: An integrated automated system for analysing evolutionary processes and generating AI-based diagnostic tools for microbes.

Mfuphi, N.*

University of Pretoria

The project aims to develop an automated analytical system for evolutionary analysis and AI-based diagnostic tools. This system will store diverse data types, such as sequences, polymorphisms, phylogenetic trees, and metadata, while preserving a tree-like data structure. It is focused on analyzing evolutionary processes like antibiotic resistance development. Additionally, the system is designed to generate AI models and diagnostic tools to address user queries regarding taxonomic identification, antibiotic resistance prediction, and more. The workflow involves using program scripts to create AI models with publicly available Python modules, such as Pandas and Keras. The designed program automatically generates software tools, allowing user access to these models and improving data parsing tools for efficient processing. The system utilizes a database structure and offers a command-line interface for user interaction. The project's current implementation is under testing and will later be available for download from GitHub. The system's capabilities include navigating tree structures, accessing allelic states, and creating AI models based on selected polymorphic sites. Future developments will focus on refining AI model generation methods and optimizing performance for large datasets, such as the Mycobacterium tuberculosis (Mtb) database of polymorphic sites, suitable for lineage identification and antibiotic resistance prediction. The developed tools for predicting Mtb pathogenicity will not only be of practical importance but will also serve as a proof of concept for the application of AI approaches to solve various practical questions in medical microbiology and biotechnology.