(15e) Proteomics Based Multivariate Random Forest Method for Prediction of Protein Separation Behavior During Downstream Purification

Proteomics
based multivariate random forest method for prediction of protein separation
behavior during downstream purification

Ryan K. Swanson¹, Ruo
Xu², Dan Nettleton² and Charles E. Glatz¹

¹Department of Chemical and
Biological Engineering, Iowa State University, Ames, IA 50011

²Department of Statistics, Iowa
State University, Ames, IA 50011

Optimizing the downstream
purification process remains a significant challenge when using biological
expression hosts for the production of recombinant proteins. An important
reason why is due to the lack of knowledge of the host cell proteins' (HCP)
separation behavior during the selected purification methods. The work to be
presented addresses this issue by modeling the separation behavior of a HCP
extract from an arbitrary expression host (corn) during commonly used
chromatographic and non-chromatographic methods. Aqueous two phase system
(ATPS) partitioning followed by two dimensional electrophoresis (2DE) provided
data on the three physicochemical properties most commonly exploited during
downstream purification for each HCP; isoelectric point, molecular weight and
surface hydrophobicity. A multivariate random forest (MVRF) statistical methodology
was then applied to this database of characterized proteins creating an accurate
tool for predicting the separation behavior of a mixture of proteins for three
commonly selected purification methods; cation exchange chromatography (CEX),
hydrophobic interaction chromatography (HIC) and ammonium sulfate precipitation
(ASP). The next logical goal will be to link these three methods in series
(i.e. ASP followed by HIC followed by CEX) and test the prediction accuracy of
the MVRF methodology during this simulated downstream purification process. Also
discussed will be the future capabilities of this work including the ability to
optimize a downstream purification process without requiring any initial
product or time spent screening potential methods in the lab regardless of expression
host.