Diagnostics of viral infections using high-throughput genome sequencing data

Area of Science:

• Plant Pathology
• Bioinformatics
• Genomics

Background:

• Plant viral infections cause substantial global economic losses, estimated at $350 billion USD in 2021.
• Effective disease management hinges on accurate and efficient viral diagnostics, as no treatments exist for infected plants.
• High-throughput sequencing (HTS) offers a cost-effective method for identifying known and novel plant viruses.

Purpose of the Study:

• To develop an automated analysis pipeline (IIMI) for diagnosing plant viral infections using HTS data.
• To address limitations in existing diagnostic methods, including subjective parameter selection and artifact-induced inaccuracies.
• To improve the accuracy, robustness, and efficiency of plant virus diagnostics.

Main Methods:

• Developed IIMI, a machine learning-based automated pipeline for diagnosing infections from 1583 plant viruses using HTS data.
• Implemented a data-driven approach for parameter selection to minimize subjectivity.
• Integrated automated filtering of sequence data regions affected by artifacts to enhance diagnostic precision.

Main Results:

• IIMI demonstrated superior performance compared to existing methods in tests using both in-house and published datasets.
• The pipeline successfully reduces subjectivity in parameter selection and effectively filters out artifacts.
• IIMI provides a prediction model and valuable resources on plant virus genomes, including artifact-prone regions.

Conclusions:

• IIMI offers a more accurate, automated, and robust solution for plant virus diagnosis from HTS data.
• The pipeline's data-driven parameter selection and artifact filtering significantly improve diagnostic outcomes.
• IIMI enhances accessibility through an R package and planned integration with the Virtool web application.