ASPIRE (Algorithms for Single Particle Reconstruction)

The original ASPIRE package was developed to prototype new ideas on algorithms for single particle reconstruction. It consists of more than 100,000 lines of Matlab code and includes many advanced algorithms in different aspects of CryoEM 3D map reconstruction such as fully automatic particle picking, denoising and classification of 2D images, 3D orientation estimation at large noise, covariance 3D analysis for structure variability and molecular heterogeneity. I am the main developer of the new ASPIRE Python package targeting a reusable and sustainable package for internal and external users and developers.

Academic IMPACT Molecular Simulation Program

IMPACT is a molecular mechanics and molecular dynamics simulation engine developed starting in the 1980's in Ron Levy's group at Rutgers University. Joint development with Schrodinger Inc. started in 2000 and it is one of basic components for many other products for structure-based drug discovery. I was the key developer of academic version of IMPACT in Ron Levy group at Temple University, with three major goals: 1) including more advanced sampling methods; 2) enabling GPU computing with openMM library; 3) porting IMPACT as a MD engine onto IBM World Community Grid for distributed computing environments.

Asynchronous Replica Exchange (AsyncRE) Framework

Traditional synchronous replica exchange methods prefer stable, tightly coupled, and homogeneous computing architectures due to the centralized synchronization steps for exchanging thermodynamic states after a MD period. In contrast, the AsyncRE framework we developed at Temple University removes the synchronizing concept and was designed to facilitate large-scale or massive-scale REMD simulations on distributed grid computing environments such as heterogeneous high performance clusters, local campus grids, and massive IBM World Community Grid.

BEDAM for Estimating Absolute Binding Free Energy

The Binding Energy Distribution Analysis Method (BEDAM) estimates absolute binding free energies for protein-ligand systems, based on a statistical mechanics theory of molecular association, parallel Hamiltonian replica exchange method, AGBNP implicit solvation model, and reweighting technique of multi-state statistical inference. I was the major developer for more advanced BEDAM methods using biasing potentials and AsyncRE framework, and Python workflows to prepare, submit, and post-analyze massive BEDAM simulations using the commercial and academic IMPACT.

Other Software Packages

I also developed in-house software packages for my postdoc and graduate projects such as NMR calculations, structure ensemble predictions of complex carbohydrates, simulations of earthquake models, and simulations of liquid materials.