Contents

BMB 961 Sec 003: Machine Learning for Molecular Dynamics

BMB 961 Sec 003: Machine Learning for Molecular Dynamics

Course Description

For almost 50 years, researchers have used in silico approaches to gain insight into the dynamics of molecular systems. Advances in computational hardware and algorithms have greatly increased the reach of these simulations both in the size of the systems studied and the timescales of the events of interest. Machine learning is bringing transformative changes to the field of molecular dynamics, impacting how initial structures and parameters are generated, how simulations are run, and how they are analyzed. This graduate-level course will provide an introduction to both machine learning and molecular dynamics theory. Students will learn about modern tools for both MD and ML, and apply them to problems in a collaborative, hands-on environment.

By the end of this course, you will be able to:

  1. Have a working understanding of the basic algorithms used in machine learning and molecular dynamics.

  2. Setup, run and analyze molecular dynamics simulations for a system of interest.

  3. Build machine learning models and apply them to prediction problems.

  4. Apply machine learning models to a variety of molecular dynamics problems.

We will work toward these goals by learning material through readings and lectures and applying this knowledge in hands-on coding laboratories, where programs will be written both individually and in small groups. The course will culminate in a research project where students will apply these skills to study a problem of their own choosing.

Topics covered

A preliminary list of topics covered in this course are listed below. Specific libraries or tools are listed in bold.

  • Repositories, packages and Google Colab

  • Introduction to Machine Learning (pytorch)

  • Hyperparameter optimization (skorch)

  • Advanced ML architectures

  • Structure prediction (AlphaFold)

  • Molecular dynamics simulation

  • Biomolecular system preparation (CHARMM-GUI)

  • Biomolecular simulation (OpenMM)

  • Analysis of MD simulations

  • ML for MD parameter generation

  • ML integrators for MD

  • Analysis of MD using neural networks (VAMPnets)

Required reading materials

This class has no required book or course pack. Course content will be distributed via JupyterBook and can be accessed by the following link: https://adicksonlab.github.io/ml4md-jb/intro.html. From time to time we will direct you toward outside online resources, but the main materials will be on the course website.

Class format

We plan to run this class in a hybrid format, where attendance is possible either in-person or via Zoom, both for the lecture and the laboratory components. This plan is subject to change following institutional or government recommendations. This course will meet on Tues 3:00-3:50 PM in BPS 1420 for lecture and Thurs 2:00-4:00 PM in BMB 202 for lab.

Required materials for class

Programming assignments are a critical part of the learning process in this course. To that end, if you attend in person you are expected to bring your laptop and its power cord to every class. If you are attending remotely you would need:

  • The ability to run the Zoom video conferencing software, which you can download here: https://msu.zoom.us/support/download

  • A computer with a reliable internet connection and a functional webcam, microphone and speakers.

Details regarding the software needed for this course are provided in the “Software Setup Guide” on the course website.

COVID-19 policy

As a reminder, the university has put in place a mask mandate and students are required to wear properly fitting masks during indoor class meetings. You should refrain from eating or drinking during class to avoid having to remove your mask. If you do consume food or drinks inside, you should remove the mask only to take a sip of beverage or a bite to eat, and you must replace the mask properly between each bite and sip. If you do not comply with this mask mandate, you may be asked to leave the building. If you forgot your mask, you will be allowed to leave to go get one.

If you have to miss class due to illness or self-isolation (as per the CDC recommended guidelines), your instructor will work to provide the necessary accommodations to ensure that your performance in class is not significantly impacted. However, should you find that your overall success in your courses is significantly impacted by any illness, please refer to the University policy on medical leave and withdrawal.

Inclusive classroom behavior

Respectful and responsible behavior is expected at all times, which includes not interrupting other students, turning your cell phone off, refraining from non-course-related use of electronic devices, and not using offensive or demeaning language in our discussions. Flagrant or repeated violations of this expectation may result in ejection from the classroom, grade-related penalties, and/or involvement of the university Ombudsperson. In particular, behaviors that could be considered discriminatory or harassing, or unwanted sexual attention, will not be tolerated and will be immediately reported to the appropriate MSU office (which may include the MSU Police Department).

In addition, MSU welcomes a full spectrum of experiences, viewpoints, and intellectual approaches because they enrich the conversation, even as they challenge us to think differently and grow. However, we believe that expressions and actions that demean individuals or groups comprise the environment for intellectual growth and undermine the social fabric on which the community is based. These demeaning behaviors are not welcome in this classroom.

Accommodations

If you have a university-documented learning difficulty or require other accommodations, please provide me with your VISA as soon as possible and speak with me about how I can assist you in your learning. If you do not have a VISA but have been documented with a learning difficulty or other problems for which you may still require accommodation, please contact MSU’s Resource Center for People with Disabilities (355-9642) in order to acquire current documentation.

Instructor contact information

Course Instructors:

Alex Dickson
alexrd@msu.edu
Associate Professor, Dept of Biochemistry & Molecular Biology and Dept of Computational Mathematics, Science & Engineering

Michael Feig
feig@msu.edu
Professor, Dept of Biochemistry & Molecular Biology

Office hours are available by appointment.

Grading information

The grading breakdown for this course is as follows:

Grade Item

Points

Participation, attendance, in-class activities

50

Independent study proposal

10

Independent study project

40

Total

100

Total: 100%

Grading scale

4.0 ≥ 90%
3.5 ≥ 85%
3.0 ≥ 80%
2.5 ≥ 75%
2.0 ≥ 70%
1.5 ≥ 65%
1.0 ≥ 60%
0.0 < 60%

previous

Welcome to Machine Learning for Molecular Dynamics - BMB 961 Sec 003

next

BMB 961 Sec 003 Semester Schedule