Learning path 02 · Beginner to advanced
Explore structures. Design proteins.
Begin by reading experimental structures and assessing their quality. Prediction, molecular simulation, and design are connected paths with different assumptions and validation needs.
Put it into practiceRead your first protein structure →BeginnerStructural biology and structure literacy
Read a structure as evidence about a molecule and an experiment.
Before you begin: Protein sequence and molecular biology basics.
What to understand
- Protein architecture, interactions, thermodynamics, and structure–function relationships.
- Experimental method, resolution where applicable, missing residues, and validation reports.
- Chains, biological assemblies, sequence mapping, and molecular visualization.
What you’ll be able to do
Inspect a structure and write a short account of what it supports and what remains uncertain.
Try this
Complete the protein-structure starter project, then locate a missing region or a validation concern in a second entry.
Keep in mind
A deposited coordinate model is an interpretation of evidence. Check the experimental method and validation information before drawing conclusions.
A few good resources
- PDB-101: understanding PDB data ↗
A welcoming introduction to structure files, experimental information, assemblies, and interpretation.
Guide · Beginner · Free · English - RCSB PDB: learning resources ↗
Find experimental structures and inspect entry-specific methods, sequence information, and validation evidence.
Database · Beginner · Free · English - UCSF ChimeraX tutorials ↗
Learn molecular visualization through worked examples; check the current software license for your intended use.
Tutorial · Beginner · Tutorials free; software terms vary by use · English - Biopython Tutorial and Cookbook ↗
Practical biological sequence and structure handling, including parsers used in the protein-structure starter project.
Documentation · Intermediate · Free · English
IntermediateStructure prediction and confidence
Understand the question a prediction answers and how to assess its reliability.
Before you begin: Structure literacy and sequence alignment basics.
What to understand
- Sequence-to-structure prediction, templates, alignments, and complexes.
- Local confidence, relative-domain confidence, and model-specific quality measures.
- Sequence correspondence, experimental comparisons, and sources of uncertainty.
What you’ll be able to do
Explain why a predicted region or interface is more or less trustworthy.
Try this
Choose a protein with both experimental and predicted structures. Align the sequences first, inspect confidence, and document which regions are comparable.
Keep in mind
Confidence scores are not interchangeable with experimental B-factors. A confident prediction alone does not establish binding, function, or a mechanism.
A few good resources
- EMBL-EBI: AlphaFold practical guide ↗
Learn access routes, interpretation, strengths, limitations, and confidence assessment for predicted structures.
Course · Intermediate · Free · English - Interpreting AlphaFold confidence scores ↗
Understand local confidence and relative-domain uncertainty before using a prediction to support a biological claim.
Guide · Intermediate · Free · English - ColabFold ↗
An accessible prediction toolkit with examples and execution options. Check current compute availability before planning a run.
Tool · Intermediate · Code free; compute access varies · English - PDB-101: understanding PDB data ↗
A welcoming introduction to structure files, experimental information, assemblies, and interpretation.
Guide · Beginner · Free · English
AdvancedProtein and binder design
Follow a design problem from a biological objective to candidates that need experimental testing.
Before you begin: Structure prediction and assessment, protein interactions, and basic scripting.
What to understand
- Target preparation, biological assembly, accessible sites, and design constraints.
- Backbone generation, sequence design, structure prediction, and candidate filtering.
- Compare diffusion-based backbone design with optimization-based pipelines such as BindCraft.
- Diversity, controls, expression and solubility considerations, and experimental validation.
What you’ll be able to do
Describe a candidate-selection strategy and the evidence needed to test a proposed binder.
Try this
Read one published design study and map its stages, selection metrics, controls, and experimental outcomes. Begin with published outputs before scheduling expensive generation runs.
Keep in mind
A favorable computational score is a screening signal. Report it separately from measured affinity, specificity, and experimental success.
A few good resources
- RFdiffusion ↗
Explore diffusion-based protein backbone generation and conditional design through the authors' code and examples.
Tool · Advanced · Code available; GPU compute may cost · English - ProteinMPNN ↗
Study sequence design conditioned on a protein backbone as a distinct stage in a broader design workflow.
Tool · Advanced · Code available; compute may cost · English - EMBL-EBI: AlphaFold practical guide ↗
Learn access routes, interpretation, strengths, limitations, and confidence assessment for predicted structures.
Course · Intermediate · Free · English - BindCraft ↗
An optimization-based binder-design pipeline combining AlphaFold-based design, sequence design, and filtering. Compare its strategy with diffusion-based approaches and inspect dependency terms.
Tool · Advanced · Code available; dependency licenses and GPU costs vary · English
AdvancedDiffusion models in molecular design
Connect denoising and conditional generation to a biological design task.
Before you begin: Basic machine learning; probability and linear algebra help with the deeper material.
What to understand
- Forward corruption, learned denoising, sampling, and conditioning.
- Representation choices, geometric constraints, and symmetry.
- Training versus inference, diversity, and evaluation beyond visually plausible structures.
What you’ll be able to do
Explain what is generated, what conditions the generation, and how candidates are evaluated.
Try this
Draw the sampling process for a binder-design example and identify where target information and constraints enter. Continue with the AI foundations path for the mathematics.
Keep in mind
Learn the model's task and representation explicitly. Different diffusion systems do not share identical inputs, outputs, or validation criteria.
A few good resources
- RFdiffusion ↗
Explore diffusion-based protein backbone generation and conditional design through the authors' code and examples.
Tool · Advanced · Code available; GPU compute may cost · English - Dive into Deep Learning ↗
Build conceptual and practical foundations in deep learning; select chapters relevant to the model you are studying.
Book · Intermediate · Free online · English
AdvancedMolecular simulation and biophysics
Explore molecular motion through models with explicit physical assumptions.
Before you begin: Thermodynamics, structural biology, and command-line basics.
What to understand
- Force fields, parameterization, solvation, and boundary conditions.
- Minimization, equilibration, sampling, replicates, and convergence.
- Trajectory analysis and the relationship between observables and a biological question.
What you’ll be able to do
Explain the assumptions and sampling limitations of a simulation study.
Try this
Follow an introductory simulation example and record the force field, ensemble, integration settings, and analysis choices.
Keep in mind
A stable-looking trajectory alone does not demonstrate convergence or establish a biological mechanism.
A few good resources
- OpenMM user guide ↗
Learn simulation setup, execution, and analysis with explicit attention to the physical model and settings.
Documentation · Advanced · Free; compute may cost · English - GROMACS documentation ↗
A reference for molecular dynamics methods, file formats, commands, and reproducible simulation configuration.
Documentation · Advanced · Free; compute may cost · English