Receptors have previously been computationally designed to bind new cognate ligands selectively to closely related molecules and have been shown to have affinities surpassing their naturally evolved parent proteins. In this work, we compiled a comprehensive dataset of previous protein engineering studies that changed cofactor specificity from NADP(H)→NAD(H), and NAD(H)→NADP(H). We analyzed the metrics of volume, hydrophobicity, and charge, and rationalized their effects on the cofactor affinities for different mutations on a cross-section of that dataset. Calculated cofactor binding energy was also explored and verified as a surrogate for computational cofactor alteration. The implicit solvation models generalized born with molecular volume and generalized born with simple switching were integrated in the Iterative Protein Redesign and Optimization framework to redesign Candida boidinii xylose reductase (CbXR) to use NADH as its cofactor by finding the optimal set of mutations in the CbXR binding pocket. NAD(H) is an order of magnitude less expensive than NADP(H), much more stable, and its concentration is roughly an order of magnitude greater than that of NADP(H). IPRO generated five libraries utilizing various tolerances in the energy cutoff and metropolis criterion. The modified version of IPRO was also used to predict mutations for different design positions of the bacterial transcriptional regulatory protein, AraC to bind different conformations of D-arabinose. This computational framework was conducted in coordination with an experimental study that successfully redesigned the AraC protein to have altered effector selectivity capable of distinguishing between different forms of the arabinose sugars.