Knowles' paper discussed how using recognizable catalytic devices could theoretically lead to the design of new ones.  The reaction catalyzed by triosephosphate isomerase (TIM), which stabilizes the intermediate in the conversion of DHAP to GAP via a 'lid', was assessed with a wild-type versus a mutant of TIM.  An organic base was used as a control to compare the rate of the reaction.  This paper demonstrated some of the catalytic features of enzymes that have been previously understood, and exposed some undiscovered areas. This mechanism of catalysis showed that alterations in the positions of residues could significantly disrupt the catalytic power of the enzyme.  This demonstrated that constructing an active site is incredibly difficult because we can predict the type of residues, but not necessarily their order and location.

Dwyer et al.'s experiment mutated the active site of a ribosome-binding protein (RBP) to exhibit TIM activity.  Computation design techniques were used to predict the mutations for the RBP.  To design the catalytic activity, they placed reactive amino acid sides chains within an active site in geometries compatible with elements of the substrate-enzyme reaction.  They focused on favorable stereochemistry and not reactivity for the rest of the side chain.  A series of mutations were performed to achieve activity.  This paper showed that although enzymes could be designed by a computational approach, there were still some drawbacks.  For instance, conclusive results could not be obtained from the crystal structures because they were all unstable.  Secondly, all of the images in the paper were computer-generated, without the use of a valid source like XRAY crystallography.  Thirdly, the catalytic activity reported was microscopic compared to that of the wild-type, thus being catalytically insignificant.

Allert et al's paper reported on the computational design of soluble protein receptors for pinacolyl methyl phosphonic acid (PMPA).  Computational design techniques allowed ligand-binding pockets of GBPs and RBPs to bind to PMPA instead of their cognate sugars, which resemble PMPA. More than half of the designs tested exhibited PMPA-dependent changes in emission intensity of a fluorescent reporter.  The results demonstrated that the designed receptors could be successfully predicted, however, the number of possible solutions to the design problem was numerous.  The main limitation was the thermostability of the receptors. They were simply too weak to be practical.  This was on the other hand a significant step towards a method of creating fluorescent biosensors.

Knowles suggested that by knowing all aspects of a catalytic mechanism, computational design of proteins was possible, but very difficult based on the structural precision needed for significant activity.  The other two papers tested this hypothesis with two different methods.  Dwyer showed that building a catalytic active site was very difficult due to the precision needed for the whole protein and not just the residues reacting in the active site.  Allert demonstrated that the computational design for mutating the active site to bind to specific molecules was possible yet did not signify the creation of a complete catalytic active site.

The significance of these papers is that the creation of a complete protein is not yet possible. However, they do emphasize that more experiments using computational protein design will enrich our knowledge of the way an enzyme is put together. By combining our knowledge of protein creation with computational protein design we will be able to create new proteins. It is pretty obvious that being able to create new proteins from scratch would be ubiquitous in all fields and significantly advance the human race.