📚 Syllabus & Course Curriculum
Bioinformatics & Computational Biology
Module-by-module breakdown of Treat DNA Like Code: Transformer Models for De Novo DNA Sequence Optimization, from foundations to a certified capstone project.
Design Space
• Codon usage, GC content, secondary structure and repeats as design constraints
• Expression, stability and manufacturability as competing objectives
• Why a sequence that scores well computationally can fail in the organism
Generation
• Autoregressive generation, masked infilling and diffusion over sequences
• Conditioning on function, host organism or expression target
• Sampling temperature and the diversity-against-validity trade-off
Optimisation
• Codon optimisation beyond frequency tables, including harmonisation
• Promoter, RBS and UTR design for a target expression level
• Guided generation with an oracle model, and reward hacking of that oracle
Filtering
• Synthesis constraints: repeats, homopolymers, extreme GC and forbidden sites
• Off-target and toxicity screening before ordering
• Ranking a design library rather than committing to a single sequence
Validation
• Design-build-test-learn cycles and library scale that makes learning possible
• Measuring designs by reporter assay or sequencing-based readout
• Feeding results back as training data, and biosecurity screening obligations
Earn government-registered certification in Treat DNA Like Code: Transformer Models for De Novo DNA Sequence Optimization
e-Certificate and e-Marksheet issued on successful completion.
