Search:
Match:
1 results

Analysis

This paper introduces VAMP-Net, a novel machine learning framework for predicting drug resistance in Mycobacterium tuberculosis (MTB). It addresses the challenges of complex genetic interactions and variable data quality by combining a Set Attention Transformer for capturing epistatic interactions and a 1D CNN for analyzing data quality metrics. The multi-path architecture achieves high accuracy and AUC scores, demonstrating superior performance compared to baseline models. The framework's interpretability, through attention weight analysis and integrated gradients, allows for understanding of both genetic causality and the influence of data quality, making it a significant contribution to clinical genomics.
Reference

The multi-path architecture achieves superior performance over baseline CNN and MLP models, with accuracy exceeding 95% and AUC around 97% for Rifampicin (RIF) and Rifabutin (RFB) resistance prediction.