This article studied applicabilities of Artificial Neural Networks (ANN), aiming at partitioning of
Asparaginase (ASP). The first step investigated the partitioning of ASP using ATPS, consisting of
polyethylene glycol and different inorganic salts. The studies of ASP partition were conducted by
measuring the amounts of ASP in the upper and lower phase, followed by calculating the recovery.
PEG of 6000 and (NH4)2SO4 provided the best performance with recovery yield of 86 %. The
partitioning behavior of the approach was investigated in the presence of an impurity kind of
amylase. Related partition coefficient of 0.21 and recovery yield of ASP of 86 % were calculated.
A Circular Dichroism (CD) spectroscopy compared the structure of standard and partitioned ASP,
which remained stable. The relative activity was determined to be 94 %. A feed-forward neural
network was applied to predict the best model to partition of ASP.