Comparative RSM and ANN modeling for sustainable production and purification of fungal protease using agro-industrial residues under solid-state fermentation and its industrial applications.

Proteases make up about 60% of the enzyme market, making them one of the most significant and vital classes of commercial and industrial enzymes. In this study, six fungal strains (Aspergillus oryzae, A. fumigatus, A. niger, A. flavus, A. terreus, and Trichoderma harzianum) were evaluated for their potential for protease production using solid-state fermentation (SSF). Screening of the fungal strains and low-cost substrates (wheat bran and banana peels) revealed that A. fumigatus and A. oryzae had the best protease-producing ability in terms of quality and quantity, while wheat bran supported the highest level of protease production. Response surface methodology (RSM) and artificial neural network (ANN) were used to optimize production parameters, including temperature, inoculum size, pH, moisture content, and substrate concentration. The maximum protease activity (228.2 U/mL) was achieved at pH 7.0, 30 °C temperature, 3.0 mL inoculum, 50% moisture content, and 5.0 g substrate concentration. The model was statistically significant (p < 0.05) with coefficients of determination (R²) of 93.15% for RSM and 94.65% for ANN. Stepwise purification steps yielded maximum enzyme recovery at 80% ammonium sulfate saturation, increasing protease activity to 364 U/mL, with a 3.01-fold purification by gel chromatography. SDS-PAGE analysis confirmed enrichment of protease in the purified fraction. The purified protease exhibited maximum relative activity at 40 °C and pH 8. Its activity was enhanced in the presence of Ca