In this day and age of ‘Big Data’ there has been an ever-increasing demand in the realm of life sciences, specifically in chemical and structural biology to predict, characterize the functional and structural properties of proteins, and their interactions from their sequence utilizing computational tools. With the assistance of in silico methods, carefully designed experiments may become more target-oriented and cost-effective investments. Human thrombin, one of the most extensively studied and characterized of all enzymes, and its vertebrate homologs were analyzed in the present work based on sequence and structure data obtained from UniProt and the Protein Data Bank databases. Three types of predicted parameters: intrinsic disorder, conformational heterogeneity, and variance of conformational transitions constituted the basis for the assessment of dynamic properties of catalytic sites and overall proteins in various taxonomic classes. Investigation of the catalytic center’s parameter values at the residue level reveals the dynamic aspect of the chemical mechanism of serine protease catalysis. Small molecule ligand-binding and non-binding residues are distinguished in terms of disorder and their range of conformational transitions. Comparison of the dynamic behavior of interacting residues in wild type and mutant thrombins complemented with solvent accessible surface area calculations indicates possible increase in binding affinity. The used programs accurately assess the dynamic properties of overall proteins and regions, although they are not specific enough for the identification of new binding sites. The calculations altogether demonstrated the dynamic nature behind the functional diversity of thrombin reported in the scientific literature over the last decade. The promising results imply that with precisely annotated training datasets of small molecule ligand-binding residues, including information from surface area calculations the algorithms could be significantly improved.