01 — Industrial biotech
Screening twelve strains on a four-run budget
- Candidates evaluated
- 12
- Runs required
- 4
- Predicted yield spread
- 2.3×
The problem
Twelve candidate production strains and funding for four shake-flask runs. The team needed a defensible way to choose which four, and what to conclude if none performed.
The work
Yield and growth parameters were estimated from published data, then a sensitivity run identified which variables the ranking actually depended on. A staged decision tree was built so early readouts would eliminate candidates instead of merely ranking them.
Delivered
A four-run screening plan with predicted spread between candidates, the two variables to hold constant, and a pre-agreed stopping rule.