Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Implementation of a biomolecule structure solution pipeline based on AlphaFold prediction

Fastus, Dominique (2024) BINP50 20231
Degree Projects in Bioinformatics
Popular Abstract
Enhancing protein structure solutions in experiments with AI

Machine learning approaches are gaining in popularity to solve previous complex and difficult problems with an ease. AlphaFold is one representative example, by using machine and deep learning to predict three-dimensional structures of biological molecules, mainly proteins, with high accuracy. Because the predicted structures are often similar to experimentally determined structures, AlphaFold models can be used to accelerate the reconstruction of the structure from X-ray diffraction data.

In X-ray crystallography, a crystal containing the protein of interest is shot with an intense X-ray beam, e.g. generated by synchrotron radiation. The diffracted X-rays are detected and... (More)
Enhancing protein structure solutions in experiments with AI

Machine learning approaches are gaining in popularity to solve previous complex and difficult problems with an ease. AlphaFold is one representative example, by using machine and deep learning to predict three-dimensional structures of biological molecules, mainly proteins, with high accuracy. Because the predicted structures are often similar to experimentally determined structures, AlphaFold models can be used to accelerate the reconstruction of the structure from X-ray diffraction data.

In X-ray crystallography, a crystal containing the protein of interest is shot with an intense X-ray beam, e.g. generated by synchrotron radiation. The diffracted X-rays are detected and provide positional information about the atoms within the crystal, which is then used to rebuild a computational representation of the protein. Lund's very own MAX-IV synchrotron facility is a key player in this field, attracting researchers and companies keen to explore new molecules for potential pharmaceutical and other applications. However, like all crystallography methods, experiments at MAX-IV face the challenge of not being able to directly obtain the protein's structure from experimental data due to information loss during detection. Predicted models from AlphaFold can provide the missing information, leading to reliable experimentally determined structures

The project further simplified this problem by directly implementing the deep learning method into the software infrastructure at the MAX-IV synchrotron research facility. Previously developed computer programs were optimized and implemented with AlphaFold to develop a user-friendly, scalable, and easy to maintain pipeline, where the final experimental structure is just one button away. This should have a substantial impact, as well as an improvement on further standardized and automatization in structure biology research

Master’s Degree Project in Bioinformatics 30 credits 2024 Department of Biology, Lund University

Advisor: Ana Gonzalez MAX IV Laboratory, Lund University, Lund, Sweden (Less)
Please use this url to cite or link to this publication:
author
Fastus, Dominique
supervisor
organization
course
BINP50 20231
year
type
H2 - Master's Degree (Two Years)
subject
language
English
id
9177899
date added to LUP
2024-11-14 15:10:21
date last changed
2024-11-14 15:10:21
@misc{9177899,
  author       = {{Fastus, Dominique}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Implementation of a biomolecule structure solution pipeline based on AlphaFold prediction}},
  year         = {{2024}},
}