Short answers, pulled from the story.
Irwin Kuntz led the team at the University of California, San Francisco that created the first software capable of predicting how small molecules bind to proteins. This group established a new field of computational chemistry and developed the initial version of UCSF DOCK.
The initial version of DOCK used geometric algorithms that placed spheres within the protein pocket to identify potential binding sites. The program performed bipartite matching to align the ligand with those points and used spheres to map the binding site.
The anchor and grow algorithm was introduced in versions four through six of the software to handle the dynamic nature of molecules. This method allowed the program to simulate the movement of ligands by identifying a stable core and growing the rest of the molecule around it.
David A. Case's group implemented a molecular dynamics engine into version six of the software and introduced the AMBER score to the scoring function. This addition allowed the program to account for receptor flexibility and rank order molecules based on energetic ensembles.
Brian K. Shoichet, David A. Case, and Robert C. Rizzo were among the codevelopers who helped shape the program into a powerful tool for molecular modeling. Shoichet focused on the application of the software to drug discovery while Case contributed to the implementation of molecular dynamics.