This project addressed the challenge of high failure rates and costs associated with the experimental validation of SSO probes. I developed and trained a supervised machine learning model by leveraging a large dataset of historical Quality Control (QC) records. The model was trained to identify complex patterns within a probe’s DNA sequence that correlate to its real-world hybridization performance. As a result, this predictive tool can now accurately score new probe designs in silico, enabling the pre-emptive filtering of poor candidates and streamlining the development of more reliable and cost-effective diagnostic kits.
Key Skills & Technologies Demonstrated#
- Machine Learning: Supervised Learning, Predictive Modeling, Model Training & Validation
- Data Science: Data Preprocessing, Feature Engineering from Biological Data, QC Data Analysis
- Bioinformatics: DNA Sequence Analysis, In Silico Modeling, Hybridization Principles
- Reduced R&D Costs: Reduces experimental synthesis and testing.
- Transformed Historical Data into a Predictive Asset: Demonstrated the ability to unlock value from archived QC data, turning a static record of past performance into a dynamic tool for future prediction.