Assessment of Alphafold methods in structural prediction of toxin-antitoxin systems with a focus on interface analysis
(2023) BINP51 20221Degree Projects in Bioinformatics
- Abstract
- The recent revolution in structural prediction methods has made modeling novel protein structures a feasible analysis method. In particular, the Alphafold protocol can produce accurate models with atomic resolution. This work aims to model multiple novel toxin-antitoxin systems and to understand the interactions of their components through structural prediction and interface analysis. Structural predictions for three different dimers and one trimer, for each toxin-antitoxin system, are made with Alphafold-multimer. Quality and interface metrics rank predictions they assess in an order that is expected, where biologically necessary toxin-antitoxin dimers score the highest and sub-optimal dimers with toxin-chaperone arrangement have the... (More)
- The recent revolution in structural prediction methods has made modeling novel protein structures a feasible analysis method. In particular, the Alphafold protocol can produce accurate models with atomic resolution. This work aims to model multiple novel toxin-antitoxin systems and to understand the interactions of their components through structural prediction and interface analysis. Structural predictions for three different dimers and one trimer, for each toxin-antitoxin system, are made with Alphafold-multimer. Quality and interface metrics rank predictions they assess in an order that is expected, where biologically necessary toxin-antitoxin dimers score the highest and sub-optimal dimers with toxin-chaperone arrangement have the lowest scores. This trend is extended to trimeric predictions, by producing shared interfaces between dimers and respective trimers. The toxin-antitoxin dimers share the most interfacing residues with trimers. More complex predictions are made for the toxin-antitoxin system with the best overall quality, the structures of which are consistent with results in previously conducted research. Evidence for multiple distinct conformations and binding arrangements is found in the interfaces of these predictions. In conclusion, structural predictions of toxin-antitoxin systems produce accurate structures that are plausible and comparable to true structures. Analyzing the interfaces of structural predictions gives valuable insights into the structure and interactions of the components in toxin-antitoxin systems. (Less)
- Popular Abstract
- Unfolding Toxin-Antitoxin Systems: Alphafold Predicts Structural Insights
Toxin-antitoxin (TA) systems defy the standard mechanism of toxicity, where the target is another organism. Toxins in TAs affect the organism that produces them, and neutralizing antitoxins are produced along with the toxins. These enigmatic systems have a variety of proposed functions. The most engaging purpose for TAs is the defense against phages; viruses that infect bacteria. Upon phage infection, the TA system activates by losing its antitoxin and poisoning its host bacterium, resulting in a dormant cell incapable of propagating phage infections into other bacteria. To better understand TAs, the structures of novel TA proteins are predicted using the... (More) - Unfolding Toxin-Antitoxin Systems: Alphafold Predicts Structural Insights
Toxin-antitoxin (TA) systems defy the standard mechanism of toxicity, where the target is another organism. Toxins in TAs affect the organism that produces them, and neutralizing antitoxins are produced along with the toxins. These enigmatic systems have a variety of proposed functions. The most engaging purpose for TAs is the defense against phages; viruses that infect bacteria. Upon phage infection, the TA system activates by losing its antitoxin and poisoning its host bacterium, resulting in a dormant cell incapable of propagating phage infections into other bacteria. To better understand TAs, the structures of novel TA proteins are predicted using the state-of-the-art method; Alphafold.
Structural predictions are made with Alphafold, a protocol that uses neural-network architecture to predict protein structures, accurate to the atom, from amino-acid sequences. Alphafold consists of two major components. The “biological” component finds sequences related to queries creating a profile of similar sequences. The “structural” component produces a loose atomic-backbone and joins side-structures to it. Predictions are improved iteratively by exchanging information between the components, resulting in a finalized model of protein structure.
An overall baseline for quality is anticipated and observed in predictions of TAs. The neutralization of toxins by antitoxins is necessary for the survival of the bacterium. Chaperones, which are proteins that stabilize other proteins, may also stabilize antitoxins, thereby promoting the neutralization of toxins. In TAs, toxins typically do not interact with chaperones. The quality of produced TA structural predictions should follow this ordering for them to be biologically feasible.
The predicted structures align with quality expectations across all assessment metrics. Use of Alphafold is validated in terms of predicting multiple types of TA structures. The produced structural predictions broaden understanding of TA systems and the interactions of their components. These results offer insights into designing proteins and molecules that activate TA systems, presenting an alternative means of antimicrobial therapy that does not rely on traditional antibiotics, potentially providing a solution to the growing problem of antibiotic resistance.
Master’s Degree Project in Bioinformatics 45 credits 2022
Department of Biology, Lund University
Advisor: Gemma Atkinson
Department of Experimental Medicine Lund, Protein Evolution (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9113046
- author
- Odai, Roni Jonus
- supervisor
- organization
- course
- BINP51 20221
- year
- 2023
- type
- H2 - Master's Degree (Two Years)
- subject
- language
- English
- id
- 9113046
- date added to LUP
- 2023-03-30 15:00:46
- date last changed
- 2023-03-30 15:00:46
@misc{9113046,
abstract = {{The recent revolution in structural prediction methods has made modeling novel protein structures a feasible analysis method. In particular, the Alphafold protocol can produce accurate models with atomic resolution. This work aims to model multiple novel toxin-antitoxin systems and to understand the interactions of their components through structural prediction and interface analysis. Structural predictions for three different dimers and one trimer, for each toxin-antitoxin system, are made with Alphafold-multimer. Quality and interface metrics rank predictions they assess in an order that is expected, where biologically necessary toxin-antitoxin dimers score the highest and sub-optimal dimers with toxin-chaperone arrangement have the lowest scores. This trend is extended to trimeric predictions, by producing shared interfaces between dimers and respective trimers. The toxin-antitoxin dimers share the most interfacing residues with trimers. More complex predictions are made for the toxin-antitoxin system with the best overall quality, the structures of which are consistent with results in previously conducted research. Evidence for multiple distinct conformations and binding arrangements is found in the interfaces of these predictions. In conclusion, structural predictions of toxin-antitoxin systems produce accurate structures that are plausible and comparable to true structures. Analyzing the interfaces of structural predictions gives valuable insights into the structure and interactions of the components in toxin-antitoxin systems.}},
author = {{Odai, Roni Jonus}},
language = {{eng}},
note = {{Student Paper}},
title = {{Assessment of Alphafold methods in structural prediction of toxin-antitoxin systems with a focus on interface analysis}},
year = {{2023}},
}