Statistical design in accelerated downstream processing

Bioprocess development is an unending process that takes years to commercialize and is attempted by many groups in parallel all over the world. Ironically, it’s soon superseded by improved, more efficient ones. Classical optimization methods e.g. One-factorat-a-time (OFAT) approach, borrowing, sequential removal, content swapping, etc., are not efficient, erroneous and painfully slow. Modern statistical methods prove to be quicker, involve less number of experiments and are highly efficient with better reproducibility. Statistical optimizations methods such as Response Surface Methodology (RSM), Central Composite Design (CCD), and Artificial Neural Networks (ANN) provide real-optima with results closer to the desired target recoveries. In spite of the huge potential, the adaptation of statistical methods is slow and sluggish due to the mathematics involved. In this article, we attempt Cholesterol Oxidase (COD) recovery as a model process using batch bead-milling for downstream processing optimizations using RSM.