Using machine learning to design the shape of self-propelled filaments driven by molecular motors
Bio-filaments like actin or microtubule are driven by molecular motors which exert tangential forces on their backbone. These systems can be modeled as an active polymer, a bead-spring model where each bead is endowed with an active force mimicking the effect of the molecular motor. Research has shown that distribution of molecular motors significantly affects the shape and overall dynamics of the polymer [1,2,3].
In this project, you will perform a systematic study of the effect of active force magnitude and distribution along the filament backbone on its structure and dynamics. You will use the shape-related information obtained as a function of active force distribution and bending stiffness of filaments as an input for a data-driven design of polymer shape using methods like reinforcement or active learning and Bayesian or evolution strategies. The project will involve using the high-performance computing molecular dynamics software package Hoomd-blue and developing codes for data analysis and AI algorithms. The scope of this project is not limited to simulating active polymers, but allows plenty of room for method development and data-driven design based on your input and interests.
References
- Globulelike Conformation and Enhanced Diffusion of Active Polymers, Valentino Bianco, Emanuele Locatelli, and Paolo Malgaretti, Phys. Rev. Lett. 121, 217802 (2018).
- Structure and dynamics of a self-propelled semiflexible filament, Shalabh K. Anand and Sunil P. Singh, Phys. Rev. E 98, 042501 (2018).
- Effects of inertia on conformation and dynamics of active filaments, M Fazelzadeh, E Irani, Z Mokhtari, S Jabbari-Farouji, Physical Review E 108 (2), 024606 (2023)
Contact
Dr. Sara Jabbari-Farouji s.jabbarifarouji@uva.nl at University of Amsterdam. You will be working in interdisciplinary computational soft matter lab.