Proteins with alternative folds reveal blind spots in AlphaFold-based protein structure prediction

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chakravarty, Devlina, Lee, Myeongsang, Porter, Lauren L.
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929550442627072
author Chakravarty, Devlina
Lee, Myeongsang
Porter, Lauren L.
author_facet Chakravarty, Devlina
Lee, Myeongsang
Porter, Lauren L.
contents In recent years, advances in artificial intelligence (AI) have transformed structural biology, particularly protein structure prediction. Though AI-based methods, such as AlphaFold (AF), often predict single conformations of proteins with high accuracy and confidence, predictions of alternative folds are often inaccurate, low-confidence, or simply not predicted at all. Here, we review three blind spots that alternative conformations reveal about AF-based protein structure prediction. First, proteins that assume conformations distinct from their training-set homologs can be mispredicted. Second, AF overrelies on its training set to predict alternative conformations. Third, degeneracies in pairwise representations can lead to high-confidence predictions inconsistent with experiment. These weaknesses suggest approaches to predict alternative folds more reliably.
format Preprint
id arxiv_https___arxiv_org_abs_2410_14898
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Proteins with alternative folds reveal blind spots in AlphaFold-based protein structure prediction
Chakravarty, Devlina
Lee, Myeongsang
Porter, Lauren L.
Biomolecules
In recent years, advances in artificial intelligence (AI) have transformed structural biology, particularly protein structure prediction. Though AI-based methods, such as AlphaFold (AF), often predict single conformations of proteins with high accuracy and confidence, predictions of alternative folds are often inaccurate, low-confidence, or simply not predicted at all. Here, we review three blind spots that alternative conformations reveal about AF-based protein structure prediction. First, proteins that assume conformations distinct from their training-set homologs can be mispredicted. Second, AF overrelies on its training set to predict alternative conformations. Third, degeneracies in pairwise representations can lead to high-confidence predictions inconsistent with experiment. These weaknesses suggest approaches to predict alternative folds more reliably.
title Proteins with alternative folds reveal blind spots in AlphaFold-based protein structure prediction
topic Biomolecules
url https://arxiv.org/abs/2410.14898