Michael Shirts
University of Colorado Boulder, CO·Chemistry
Computer simulations of molecules can now predict how strongly a drug candidate binds its protein target, and drug companies use them to decide which compounds to make. Yet many important targets remain too costly to simulate accurately, and modeling tools for crystals and polymers lag behind.
- Fields
Chemistry
Engineering
Biophysics
- Topics
molecular simulation
molecular dynamics
force fields
- Grants
- 4 known grants, through Aug 2030
What they work on
- Build systematically improvable force fields with the Open Force Field Initiative, optimized and validated against new thermophysical data, with statistical tests of when added complexity improves accuracy.
- Develop sampling methods, augmented with machine-learning structure prediction and run asynchronously on clusters and cloud computers, to compute binding free energies of large, flexible drug ligands.
- Use generative models such as Boltzmann generators, sampling from multiple thermodynamic states, to estimate the thermodynamics of small-molecule drug crystals and their polymorphs.
- Extend generative crystal methods to bridge force-field and quantum-mechanical ensembles and to handle systems such as hydrates, with open-source Python implementations and tutorials.
- Create the Multiscale Polymer Toolkit (MuPT), Python libraries that convert polymer models between simulation engines and resolutions and generate reproducible workflows for reacting polymers.
- Establish an Institute for Computational Molecular Science Education offering summer and winter schools, web content, and course materials in molecular simulation and machine learning.
More about the research
This research improves the physical models and sampling methods behind molecular simulation. It applies them to flexible drugs, crystalline medicines, and soft materials, and builds open software and training so others can use them.
Why it matters
More accurate, efficient simulations could help drug developers move beyond small, rigid molecules to new targets and kinds of medicines. Predicting crystal properties could make pill development faster and cheaper, and could also aid work on pigments, agrochemicals, food additives, and electronic materials. Open polymer tools could speed research on recyclable plastics, membranes, and composites.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Computational tools for new pharmaceutical paradigms
NIH · Sep 2025 – Aug 2030
Collaborative Research: Frameworks: Enabling reproducible research in soft materials with the Multiscale Polymer Toolkit (MuPT)
NSF · Oct 2024 – Sep 2027
Statistical Inference of Molecular Ensembles and Application to Crystalline Solids
NSF · Aug 2024 – Jul 2027
Closest labs
Neeraj Rai
Mississippi State University, MS·Engineering
Develop machine-learned interatomic potentials, models that predict forces between atoms, using the multi-atomic cluster expansion framework trained on first-principles molecular dynamics trajectories of zeolite systems.
Jay Ponder
Washington University in St Louis, MO·Chemistry
Develop the AMOEBA+ polarizable multipole force field for DNA and RNA, with parameters generated automatically by the Poltype2 software, to model nucleic acid structure, dynamics and interactions.
Tai-Sung Lee
Rutgers University-New Brunswick, NJ·Chemistry
Build a Python-based executing environment that connects AMBER's molecular dynamics engine, a free energy module and user workflow control, so new simulation methods can be added and prototyped quickly.
The school
University of Colorado Boulder: 36 labs on PI Pages, $718M a year on research, 447 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.