PredictION

PredictION predicts coverage depth for SARS-CoV-2 genomes sequenced on Oxford Nanopore Technologies' MinION to estimate optimal read counts and guide resource allocation for sequencing.

Key Features:

• Predictive Modeling: Uses a machine learning approach with linear regression to predict coverage depth from the number of MinION reads.
• Sequencing Resource Optimization: Estimates the optimal number of reads required to achieve >200X coverage while maintaining accurate SARS-CoV-2 lineage-clade assignments.
• Yield Maximization and Flow Cell Reuse: Predicts coverage depth to inform sequencing runtime decisions and the reuse of flow cells with remaining nanopores to maximize data yield and resource utilization.
• Performance Metrics: Reports predictive performance quantified by an R squared value of -0.98.

Scientific Applications:

• Genomic Research: Planning and executing SARS-CoV-2 sequencing experiments by estimating reads needed for target coverage and reliable lineage/clade assignment.
• Resource-Constrained Settings: Enabling cost-effective sequencing strategies through reuse of reagents and flow cells and by minimizing unnecessary sequencing effort in low-resource environments.

Methodology:

Linear regression (machine learning) model trained to predict coverage depth from Oxford Nanopore Technologies MinION read counts; model performance reported as R squared = -0.98.