Introduction to AlphaFold

Duration: 12 hours / 3 days

Date: 12, 13 and 14 of June

Time: 9:30 - 13:30 pm

Level: Beginner

Format: Presential

ROOM: Dr. Aiguader 80, Barcelona - 61.212

Registration opening date: 21 of March

Deadline for registration: 14 of April

Course Description

The aim of this course is to introduce the basic principles of AlphaFold, including its relevance, usage, and applications. We will learn how to use AlphaFold to predict the 3D structure of proteins and will explore some of its applications. The course will also discuss the current state and the ongoing research in the field, including the latest developments in the field of protein structure prediction.

Why should you attend the course?

AlphaFold is one of the most successful approaches to address the protein folding problem, which refers to the large number of proteins for which the 3D structure is unknown. Knowing the 3D structure of proteins is crucial for understanding its function and its interactions with other molecules, which can help to improve our comprehension of the underlying mechanisms of many biological processes. 

This course combines theoretical concepts together with hands-on experience using AlphaFold to predict protein structures and learn how to interpret the resulting prediction. By the end of the course, we will have a comprehensive understanding of how AlphaFold works, and the confidence to run AlphaFold yourself for predicting and manipulating your proteins of interest. Finally, we will wrap up with an open table to discuss limitations, applications, and challenges that still need to be addressed, as well as future progress that goes beyond predicting protein structures.

Prerequisites

This course has no formal prerequisites besides bringing your own laptop. The following advices aim to take the most out of the course:

• Basic knowledge on structural biology: amino acid sequence, protein composition, secondary structure of a protein. 

• Programming experience is not required for this course. 

Schedule

Day 1

• Introduction to the protein folding problem and the relevance of protein structure in cell biology. 

• Protein structure prediction: workflow and methods  

• Getting to know AlphaFold and the CASP competition  

• (Practical case)  Accessing, downloading and manipulating AlphaFold predictions using the AlphaFold Protein Structure Database  (ChimeraX required) 

Day 2

• Overview the AlphaFold algorithm and the role of deep learning to obtain reliable and accurate results  

• Measuring accuracy for AF models 

• Google Colab and ColabFold 

• (Practical case)  Protein structure prediction, visualization and analysis with AlphaFold2 (ChimeraX required) 

Day 3

• Modeling protein-protein interactions: homologous interactions and docking. 

• AlphaFold-Multimer: modeling multimeric complexes / interactions between chains.

• (Practical case) Modeling a small multimeric complex with AlphaFold-Multimer (ChimeraX required)

      Closing: 

• Current application of AlphaFold (biochemistry, drug discovery, biomedicine).

• Limitations

• Other approaches: RoseTTAFOLD, OmegaFold, OpenFold, Language models.

• Future steps: increasing accuracy for long proteins, unstructured proteins, multimeric proteins, protein-DNA interactions, mutation sensitivity, modeling bigger systems such as a nuclear pore complex or an entire cell. 

Class Pace 

We encourage the participants to do as many questions as needed during the whole course.

The hands-on practical session aims to enhance dynamic learning to consolidate the theoretical concepts behind the protein structure prediction.

The Slack channel will be used for providing instructions to successfully follow the course, for a day-to-day track on the theoretical and practical concepts discussed and as a questions & answers channel.  

References

• Jumper, John, et al. "Highly accurate protein structure prediction with AlphaFold." Nature 596.7873 (2021): 583-589.

• Evans, Richard, et al. "Protein complex prediction with AlphaFold-Multimer." BioRxiv (2021).

• Varadi, Mihaly, et al. "AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models." Nucleic acids research 50.D1 (2022): D439-D444.

• Mirdita, Milot, et al. "ColabFold: making protein folding accessible to all." Nature Methods (2022): 1-4.

Short biography 

Altair C. Hernández is a PhD candidate in the Live-cell Structural Biology laboratory at Universitat Pompeu Fabra. Previously, he did the Master’s degree in Bioinformatics for Health Science at Universitat Pompeu Fabra (2018 - 2020) and the bachelor’s in Microbiology at Universitat Autònoma de Barcelona (2013 - 2017). His research interests include structural modeling of protein complexes and image analysis from live-cell microscopy.