Data-driven AI system for learning how to run transcript assemblers

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• RNA sequencing (RNA-seq) is crucial for understanding gene expression.
• Transcript assemblers reconstruct expressed transcripts but require optimal parameter settings.
• Manual parameter tuning is time-consuming and may not yield optimal results.

Purpose of the Study:

• To introduce AutoTuneX, an AI-driven system for automatic optimization of transcript assembler parameters.
• To improve the accuracy and efficiency of transcript assembly from RNA-seq data.
• To provide a data-driven strategy for parameter selection in sequence analysis tools.

Main Methods:

• Developed AutoTuneX, a machine learning system leveraging knowledge from existing RNA-seq samples.
• Trained AutoTuneX to predict optimal parameters for transcript assemblers.
• Evaluated AutoTuneX on 1588 human RNA-seq samples using two different transcript assemblers.

Main Results:

• AutoTuneX predicted parameters that led to more accurate transcript assembly in 98% of tested samples.
• Significant improvements in Area Under the Curve (AUC) were observed, with some cases reaching up to 600% enhancement.
• The system demonstrated effective knowledge transfer to optimize parameters for unseen RNA-seq samples.

Conclusions:

• AutoTuneX offers a robust, data-driven approach to automatically optimize transcript assembler parameters.
• The system substantially enhances the accuracy of transcript assembly compared to default settings.
• AutoTuneX represents a novel strategy for optimizing the performance of bioinformatics tools in sequence analysis.