Rothlisberger group has published the framework to perform QM/MM in CPMD/GROMACS or even Coarse grain and continuum model. The framework is enhanced with fast data exchange between programs by using MPI.
A group from University College London, University of Amsterdam reported their new methods in calculating the residence times of given molecules (drug) to given proteins. They claimed their method is less computational demand than the other but with the same accuracy, which can be applied in drug discovery.
Frank Noe group from FU Berlin recently published a paper relating applying the deep reinforcement learning in comformational sampling. It is well-known that to calculate thermodynamic properties or even just understanding the structural changes, it is required thoroughly sampling the free energy landscape of given molecules. This would lead to the timescale obstacle in computational sciences, and the method by Frank Noe group expect to contribute to solve this problems.
https://chemrxiv.org/articles/Targeted_Adversarial_Learning_Optimized_Sampling/7932371
The paper introduced their newly developed Relative Principal Components Analysis to apply in biomolecular conformational changes. They showed their test case with HIV-1 protease upon binding to small drug moelcules.
Recently, Nobel Laureate Martin Karplus group published a paper about the effects of box size on the dynamics of human hemoglobin [1].
It is nothing, but recently there is a group in MaxPlanck Biophysical Chemistry tested the results and found nothing strange as the paper said in [2]. The error comes from the lack of statistics.
[1] https://elifesciences.org/articles/35560
[2] https://www.biorxiv.org/content/biorxiv/early/2019/02/28/563064.full.pdf
After receiving stem-cell transplant for replacing the white blood cells with HIV-resistant versions, the second patient has been found to be cleared out of virus.
A good textbook with practical example has been published by Aston Zhang et al.
The contents is available at http://d2l.ai
Yoshua Bengio Geoffrey Hinton and Yann LeCun have been nominated for the Turing award laureates.
A deep feedforward neural network has been used to predict distance and orientation dependent electronic coupling elements in disordered molecular materials. It shows the ability of highly accurate prediction and be a replacement of actual first principle calculation for prediction.
This is an interesting paper from Grubmuller group about benchmarking the hardware with GROMACS. The OpenCL, cross platform language, which means allowing you to compile your program in not only NVIDIA card, but also AMD card and even mobile chip or PlayStation, has allowed a lot of simulation packages including GROMACS to broaden their market. Comparing the performance to price ratio, AMD cards seem to be a good choice.
Authors perform thoroughly investigation in the phase space (P,T) of Trp-cage to check stability.
This paper takes advantage of ML to search for the better ways of design kirigami-inspired cuts into a sheet to generate the stretchable materials with metamorphic properties where 2D can transform into complex 3D. The predictions are then verified by molecular dynamics simulation.
https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.121.255304
Our paper (online 19 Mar 2021) is related to the signaling transmission between A2a receptor (belonging to the large transmembrane protein family G Protein Coupled Receptors). The findings suggest a thorough link between ligand binding pocket on GPCR to the subunits on G Protein.
Link to the paper: https://doi.org/10.1016/j.cell.2021.02.041
Link to the introduction about our paper at TokyoTech site: https://www.titech.ac.jp/news/2021/049288.html
© Tran Duy 2021