We have curated over 100 structures of class A beta-lactamase proteins, a family of enzymes that is responsible for a considerable percentage of multidrug antibiotic resistance in bacteria. Our dataset includes three ancestral reconstructions of pre-hominid beta-lactamase structures unperturbed by modern day selective pressures on antibiotic resistance. Having ancient to extant protein structures represented, we sweep over the feature space governing thermodynamic stability to identify underlying cooperativity motifs that are conserved over the evolution of this protein family. Across all proteins within our dataset, mechanical cooperativity is analyzed by a minimum Distance Constraint Model in a similar fashion to previous work [1] to obtain quantitative stability/flexibility relationships (QSFR). We create an innovative thermodynamic landscape that describes the variance in thermodynamic stability across this family of proteins based upon a model parameter space that is relevant to enzyme activity. Using clustering and machine learning techniques, we identify relevant chemo-physical constraints that each beta-lactamase structure places on function and mechanical cooperativity over a diverse range of thermodynamic conditions. Under appropriate stability conditions for functionality, we aim to elucidate why antibiotic resistance differs between members in this class, despite all members sharing similar functional sites, global dynamics, and overall structural similarity. We also take into consideration the known pharmacological metadasta for each of these proteins to better understand the binding mechanisms between these enzymes and their various substrates. Our bioinformatics approach is designed to give insight into selective pressure and possible hidden mechanisms behind substrate selectivity in modern day beta-lactamases, while revealing the relevant evolutionary factors that guide ligand selectivity to change.