AlphaFold2: A role for disordered protein prediction?

Wilson, Carter J.; Choy, Wing‐Yiu; Karttunen, Mikko

doi:10.1101/2021.09.27.461910

Cited by 10 publications

(22 citation statements)

References 75 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…AlphaFold2 pLDDT not competitive as proxy for CheZOD score disorder. It has recently been suggested that AlphaFold2's pLDDT score is a reasonable indicator for disorder (Wilson et al, 2021;Piovesan et al, 2022). By plotting pLDDT against CheZOD scores (Fig.…”

Section: Discussionmentioning

confidence: 99%

SETH predicts nuances of residue disorder from protein embeddings

Ilzhoefer

Heinzinger

Rost

2022

Preprint

View full text Add to dashboard Cite

Predictions of millions of protein 3D structures are only a few clicks away since the release of AlphaFold2 results for entire data sets. However, many proteins have so-called intrinsically disordered regions (IDRs) that do not adopt unique structures in isolation. These IDRs are associated with several diseases, including Alzheimer's Disease. We showed that the absence of reliable AlphaFold2 predictions correlated only to a limited extent with IDRs. In contrast, many expert methods predict IDRs directly and reliably by combining complex machine learning models with expert-crafted input features and evolutionary information from multiple sequence alignments. Some of these input features are not always available and computationally expensive to generate, limiting their scalability. In this work, we present the novel prediction method SETH that predicts residue disorder from embeddings generated by the protein Language Model ProtT5, which explicitly only uses single sequences as input. Thereby SETH, a relatively shallow convolutional neural network, already outperformed much more complex state-of-the-art solutions while being much faster, allowing to create predictions for the human proteome in fewer than 30 minutes on a machine with one RTX A6000 GPU with 48GB RAM. Trained on a continuous disorder scale, our method captured subtle variations in disorder, thereby providing important information beyond the binary classification of other predictors. The new method is freely publicly available at: https://github.com/DagmarIlz/SETH.

show abstract

Section: Discussionmentioning

confidence: 99%

SETH predicts nuances of residue disorder from protein embeddings

Ilzhoefer

Heinzinger

Rost

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…AF2 and RF are capable of predicting the 3D structures of input sequences with extremely high accuracy. Since the release of the source codes of both tools, discussions about their application limits and applicability have continued ( 18 – 25 ). These software tools are considered to predict homo- or hetero-complex structures as well as monomer structures, although researchers around the world are currently testing whether AF2 can predict complex structures with the same accuracy as monomeric structures.…”

Section: Introductionmentioning

confidence: 99%

A Web Server for GPCR-GPCR Interaction Pair Prediction

et al. 2022

View full text Add to dashboard Cite

The GGIP web server (https://protein.b.dendai.ac.jp/GGIP/) provides a web application for GPCR-GPCR interaction pair prediction by a support vector machine. The server accepts two sequences in the FASTA format. It responds with a prediction that the input GPCR sequence pair either interacts or not. GPCRs predicted to interact with the monomers constituting the pair are also shown when query sequences are human GPCRs. The server is simple to use. A pair of amino acid sequences in the FASTA format is pasted into the text area, a PDB ID for a template structure is selected, and then the ‘Execute’ button is clicked. The server quickly responds with a prediction result. The major advantage of this server is that it employs the GGIP software, which is presently the only method for predicting GPCR-interaction pairs. Our web server is freely available with no login requirement. In this article, we introduce some application examples of GGIP for disease-associated mutation analysis.

show abstract

“…Position-based analysis of Matrin3 disorder reveals that the region between RRM2 and ZnF2 have the highest tendency for disorder, while the region spanning the RRMs has the lowest. Considering the recent computational successes in describing the structural features and dynamics of disordered proteins ( Cino et al, 2011 ; Krzeminski et al, 2012 ; Do et al, 2014 ; Ramis et al, 2019 ; Pietrek et al, 2020 ; Wilson et al, 2021a ; Chang et al, 2021 ), molecular simulation of specific protein segments could help to more accurately characterize Matrin3, FUS, and TDP-43, by producing a structural ensemble rather than a single frame. Computational methods have also been extended to study protein aggregation propensities ( Morriss-Andrews and Shea, 2015 ), liquid-liquid phase-separation ( Paloni et al, 2020 ; Benayad et al, 2021 ), IDP-protein interactions ( Cino et al, 2013 ; Do et al, 2016 ; Karttunen et al, 2018 ), and the effects of mutation ( Vacic et al, 2012 ; Wilson et al, 2021b ) and post-translational modification ( Jin and Graeter, 2021 ; Sieradzan et al, 2021 ) on protein structure, all potential avenues for investigation of a disordered ALS protein like Matrin3.…”

Section: Structure Disorder and Llps In Matrin3mentioning

confidence: 99%

Matrin3: Disorder and ALS Pathogenesis

Salem

Wilson

Rutledge

et al. 2022

Front. Mol. Biosci.

Self Cite

View full text Add to dashboard Cite

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disorder characterized by the degeneration of both upper and lower motor neurons in the brain and spinal cord. ALS is associated with protein misfolding and inclusion formation involving RNA-binding proteins, including TAR DNA-binding protein (TDP-43) and fused in sarcoma (FUS). The 125-kDa Matrin3 is a highly conserved nuclear DNA/RNA-binding protein that is implicated in many cellular processes, including binding and stabilizing mRNA, regulating mRNA nuclear export, modulating alternative splicing, and managing chromosomal distribution. Mutations in MATR3, the gene encoding Matrin3, have been identified as causal in familial ALS (fALS). Matrin3 lacks a prion-like domain that characterizes many other ALS-associated RNA-binding proteins, including TDP-43 and FUS, however, our bioinformatics analyses and preliminary studies document that Matrin3 contains long intrinsically disordered regions that may facilitate promiscuous interactions with many proteins and may contribute to its misfolding. In addition, these disordered regions in Matrin3 undergo numerous post-translational modifications, including phosphorylation, ubiquitination and acetylation that modulate the function and misfolding of the protein. Here we discuss the disordered nature of Matrin3 and review the factors that may promote its misfolding and aggregation, two elements that might explain its role in ALS pathogenesis.

show abstract

AlphaFold2: A role for disordered protein prediction?

Cited by 10 publications

References 75 publications

SETH predicts nuances of residue disorder from protein embeddings

SETH predicts nuances of residue disorder from protein embeddings

A Web Server for GPCR-GPCR Interaction Pair Prediction

Matrin3: Disorder and ALS Pathogenesis

Contact Info

Product

Resources

About