Seminar über Theorie der Kondensierten Materie/ Weiche Materie und Statistische Physik

PW 059721

Hybrid event; PW: 359038

I shall present our recent results where we modify internal-topology of ring polymers (bead-spring model) to obtain localization of polymer segments within cylindrical confinement. We tried out a variety of topologies to establish the entropic principles which lead to localization of polymer segments. Two polymers in a mixed state under cylindrical confinement undergo segregation, and again topology plays an important part in determining the forces which lead to segregation and subsequent localization. We have used this understanding to predict the localization of loci (polymer segments) of bacterial DNA polymers, as the chromosome is replicating and segregating. It is known that some simple bacterial cells do not have the required machinery to separate their chromosomes within the cell. We have matched out model simulations results for two different bacterial chromosomes, moreover, our model simulation match data from two different experimental techniques ( HiC and FISH) which are complementary in spirit. I am extending our topology driven organization understanding in a variety of scenarios, some of which will be useful to understand more complex organization of chromosomes within more complex cells as well in more complex scenarios of multifork replication within the bacterial cell.

Molecular modeling of biomolecular assemblies exemplifies a disruptive area holding both promises and contentions. We will start with a brief story of simulating the first ever cell organelle in molecular details to find how nature has chosen survival fitness over efficiency of energy transfer as an evolutionary design (Cell, 2020). Despite such advances in exascale computing, biophysical simulation continues to grapple with handling molecular diversity. So, we will employ deep learning approaches often used in Google searches, called the inception network, to marry interaction signatures from Alphafold models and proteomics analysis with predictable patient outcomes (Cell Sys, 2022). An application will highlight how molecular modeling is used at an industrial scale to de-risk vector-based vaccines for distribution across 194 countries (Sci Adv 2021). We will conclude by seeing how transient interactions are difficult to predict, and path integrals with reinforcement learning offer a possible way to track diversity of dynamics (NeurIPS, 2022).

Hybrid meeting; ID: 814 7712 9515, Passcode: 008795