SEQdata-BEACON: a comprehensive database of sequencing performance and statistical tools for performance evaluation...

Area of Science:

• Genomics
• Bioinformatics
• Sequencing Technology

Background:

• The BGISEQ-500 sequencer, utilizing DNBSEQ technology, offers high throughput and cost-effectiveness, making it prevalent in research and clinical settings.
• Understanding its performance is crucial for process stabilization, accurate result interpretation, and efficient problem-solving.

Purpose of the Study:

• To construct a comprehensive database, SEQdata-BEACON, for accumulating and analyzing BGISEQ-500 run performance data.
• To develop a predictive model for sequencing yield based on early-stage data.

Main Methods:

• Collected performance data from 60 BGISEQ-500 instruments over six months, recording 65 metrics for 2236 sequencing runs.
• Utilized correlation matrices to cluster 52 numerical metrics into yield-quality, machine state, and calibration groups.
• Developed a linear regression model using 200 cycles of data to simulate and predict final sequencing outputs.

Main Results:

• The developed linear regression model accurately simulated final sequencing outputs with an R² of 0.97.
• The model could predict final yield as early as the 15th cycle with an R² of 0.81, validated by a test set yielding R² of 0.96.
• Metric distributions provided insights into sequencing process patterns, aiding analysis and problem identification.

Conclusions:

• SEQdata-BEACON provides an updated reference for BGISEQ-500 users to understand DNBSEQ technology, resolve issues, and optimize performance.
• The predictive model demonstrates reliability and effectiveness for early yield forecasting.
• The database and tools are accessible online to support the scientific community.