Graduation Year


Document Type




Degree Granting Department

Industrial and Management Systems Engineering

Major Professor

Susana K. Lai-Yuen, Ph.D.

Committee Member

Les Piegl, Ph.D.

Committee Member

Alfredo Cardenas, Ph.D.

Committee Member

Tapas Das, Ph.D.

Committee Member

Kimon Valavanis, Ph.D.

Committee Member

Ali Yalcin, Ph.D.


collision detection, molecular conformational search, flexible molecules, molecular stability, computational geometry, differential evolution


Modeling molecular structures is critical for understanding the principles that govern the behavior of molecules and for facilitating the exploration of potential pharmaceutical drugs and nanoscale designs. Biological molecules are flexible bodies that can adopt many different shapes (or conformations) until they reach a stable molecular state that is usually described by the minimum internal energy. A major challenge in modeling flexible molecules is the exponential explosion in computational complexity as the molecular size increases and many degrees of freedom are considered to represent the molecules' flexibility. This research work proposes a novel generic computational geometric approach called enhanced BioGeoFilter (g.eBGF) that geometrically interprets inter-atomic interactions to impose geometric constraints during molecular conformational search to reduce the time for identifying chemically-feasible conformations. Two new methods called Kinematics-Based Differential Evolution (kDE) and Biological Differential Evolution (BioDE) are also introduced to direct the molecular conformational search towards low energy (stable) conformations. The proposed kDE method kinematically describes a molecule's deformation mechanism while it uses differential evolution to minimize the inta-molecular energy. On the other hand, the proposed BioDE utilizes our developed g.eBGF data structure as a surrogate approximation model to reduce the number of exact evaluations and to speed the molecular conformational search. This research work will be extremely useful in enabling the modeling of flexible molecules and in facilitating the exploration of nanoscale designs through the virtual assembly of molecules. Our research work can also be used in areas such as molecular docking, protein folding, and nanoscale computer-aided design where rapid collision detection scheme for highly deformable objects is essential.