ML Protein Design Bootcamp 2025

Self-paced course

Build judgment across modern ML protein design workflows.

This asynchronous edition preserves the Monday–Thursday sequence of the original 2025 workshop. Move from environment setup to structure prediction, backbone generation, docking, and a binder-design capstone—producing evidence at every stage.

Choose your route Begin Monday Preview capstone
Capstone workflow from target selection through validation and final design selection.

Your progress

Evidence checkpoint tracker

Check an item only after its named artifact or decision exists. Progress is saved in this browser; it measures collected evidence, not time spent or model success.

0 / 0 required evidence checkpoints
0 optional checkpoints completed
0%

Three Ways Through the Course

Route Use it when Before the capstone Estimated active effort*
Core You want the shortest coherent learning path Complete each day’s core “Read + Do + Check” work and evidence checkpoints 16–24 hours
Full bootcamp You want the breadth of the original workshop Add all recordings, activities, optional technical deep dives, and relevant installations 35–55 hours
Reference You already have a workflow or a specific question Jump by tool, concept, or target; follow the prerequisites named on that page Variable

*Downloads, scheduler queues, and GPU runtime are additional. Compare the routes, prerequisites, and compute options in Start Here →

Course Dashboard

HPC setup path showing login node, scheduler, GPU node, and verification commands. Monday4–8 core hours + installs Tool Installation Set up HPC access, CUDA, conda environments, containers, and the core prediction and design tools. Required foundationOpen Monday -> Structure prediction confidence dashboard with a pLDDT profile and PAE heatmap. Tuesday5–7 core hours + jobs Structure Prediction Use PyMOL, AlphaFold2, and ESMFold to predict structures and interpret confidence instead of treating outputs as answers. Analysis heavyOpen Tuesday -> RFdiffusion output gallery showing unconditional, motif, hotspot, and symmetric backbone designs. Wednesday3–5 core hours + jobs Advanced Prediction and Design Compare complex prediction tools, then generate and inspect protein backbones with RFdiffusion workflows. Design workflowOpen Wednesday -> CPU versus GPU timing chart showing vectorization and workload-size effects. Thursday3–5 core hours Compute, Docking, and Reflection Connect PyTorch basics to GPU performance, docking intuition, and planning decisions for the capstone. Compute intuitionOpen Thursday -> Binder design target interface showing a target chain, hotspot residues, and a designed binder. Capstone10–25 core hours + jobs Protein Binder Design Project Choose a target, define a binding face, generate binders, design sequences, validate structures, and document the rationale behind your choices. ProjectOpen Capstone ->

Getting Started

  1. Read Start Here and choose the core, full-bootcamp, or reference route.
  2. Create the evidence-portfolio folders and record your compute/account fallback.
  3. Start with Monday’s readiness checks unless your environment is already verified.
  4. Move through the historical days in order when possible; later activities assume earlier concepts even when they use a different tool.
  5. Check a progress item only after its named log, structure, figure, comparison, or decision memo exists.
  6. Return anytime and use the resume banner to pick up where you left off.

Report an Issue / Ask a Question (GitHub account required)