DL/ML approach to predict novel antimicrobial peptide 

Multi-drug resistant pathogens are posing deadly threats to humans around the world. As commonly used drugs are not sufficient, and alternative therapeutic options are limited, the mortality rate can be very high among infected patients. Therefore, there is a pressing need for new and effective antimicrobial drugs to combat this fast-evolving group of pathogens. Therefore, antimicrobial prediction model, namely AmPEP and Deep-AmPEP30, has been proposed. A new antibacterial peptide  validated by experimental assay has been reported using our methods

Knowledge based coarse grained model for protein 

The main idea behind developing a new CG force field (CGMM) is to obtain accurate protein dynamics information compared to NMR ensemble. A CGMM model for protein was defined on pseudo atoms centred at the Cα atom position of the backbone, each representing an amino acid. The potential energy function is expressed by the sum of virtual bonded and nonbonded potentials. All potential energy function has been formulated and parameterized using the non-linear regression method from probability distributions. The probability distribution of virtual bond and nonbonded geometry has been calculated using experimental PDB structures. 

Data driven approach to infer protein function form protein dynamics

Inference of the molecular function of proteins is the fundamental task in the quest for understanding cellular processes. The difficulty arises primarily due to the lack of high-throughput experimental technique for assessing protein molecular function; computational approaches are trying hard to fill the gap. Here we propose a de novo approach to annotate protein molecular function through structural dynamics match for a pair of segments from two distinct proteins. The distance topological vector obtained from different time frames from the simulation trajectory was used to identify the mode of motion for a specific function (DynaFunc).

Interaction of drug molecules with disease related protein

Tumor necrosis factor (TNF) receptor type II (TNF2) is predominantly expressed on CD4+ Foxp3 regulatory T cells (Tregs). It plays a major role in the function of Tregs. Therefore, TNFR2 antagonist has the potential to enhance antitumor immune responses, by upregulating or downregulating Treg activity. To shed light on the potential TNFR2-targeting small molecules, we performed molecular docking and umbrella sampling molecular dynamics simulation. Our result suggested that region 3 is a potentially more viable binding site to target by small moleclues than region 4. 

Investigate the molecular mechanism of protein translocation mediated by Sec61 complex

One-third of human proteins are transported into the cell’s endoplasmic reticulum via the heteromeric Sec61 complex. The protein conducting channel (the Sec61 complex) is evolutionarily conserved. The α-subunit of the complex,Sec61α, plays a central role in the translocation of nascent polypeptides emerging from the translating ribosome. To accomplish protein translocation, the Sec61α undergoes substantial conformational changes. Those changes are associated with accessory proteins (Sec63, Sec62, TRAP etc.). Defect in protein translocation has been linked to many diseases, including cancer an hereditary human diseases (present of inhibitor or mutation). By comparing simulations of the complex (Sec complex) to those of unbound Sec61α we shed light on translocation mechanism in present of accesory protein (post-translational). 

To design efficient biosensor: refined empirical force field to model protein and self-assembled monolayer (SAM) Interactions

Understanding protein interaction with material surfaces is essential for the development of nanotechnologies such as implant biomaterials, drug delivery systems, and biosensors. The interaction of the protein with SAM needs to be adequately modelled in the biomolecular simulation. To obtain the improved force fields for simulation, we systematically tuned the Lennard-Jones parameters of selected amino acid sidechains and the functional group of SAM. The final parameter set achieved close agreement in the experimental peptide adsorption free energy with a mean square error of 0.65 kcal/mol compared to 1.5 kcal/mol using standard parameters. Furthermore, using the modified parameters, the adsorption of lysozyme on SAMs was studied.