Benchmarking AlphaMissense pathogenicity predictions against cystic fibrosis variants

PLoS ONE, Год журнала: 2024, Номер 19(1), С. e0297560 - e0297560

Опубликована: Янв. 25, 2024

Variants in the cystic fibrosis transmembrane conductance regulator gene (CFTR) result fibrosis-a lethal autosomal recessive disorder. Missense variants that alter a single amino acid CFTR protein are among most common variants, yet tools for accurately predicting molecular consequences of missense have been limited to date. AlphaMissense (AM) is new technology predicts pathogenicity based on dual learned structure and evolutionary features. Here, we evaluated ability AM predict variants. predicted high residues overall, resulting false positive rate fair classification performance CF from CFTR2.org database. score correlated modestly with metrics persons including sweat chloride level, pancreatic insufficiency rate, Pseudomonas aeruginosa infection rate. Correlation was also modest trafficking folding competency vitro. By contrast, well channel function vitro-demonstrating training approach learns important functional information despite lacking such data during training. Different across indicated may determine if polymorphisms cannot differentiate mechanistic effects or nature pathophysiology. Finally, predictions offered utility inform pharmacological response i.e., theratype. Development approaches biochemical properties therefore still needed refine targeting emerging precision therapeutics.

Язык: Английский

---

Analysis of AlphaMissense data in different protein groups and structural context

Scientific Data, Год журнала: 2024, Номер 11(1)

Опубликована: Май 14, 2024

Abstract Single amino acid substitutions can profoundly affect protein folding, dynamics, and function. The ability to discern between benign pathogenic is pivotal for therapeutic interventions research directions. Given the limitations in experimental examination of these variants, AlphaMissense has emerged as a promising predictor pathogenicity missense variants. Since heterogenous performance on different types proteins be expected, we assessed efficacy across several groups (e.g. soluble, transmembrane, mitochondrial proteins) regions intramembrane, membrane interacting, high confidence AlphaFold segments) using ClinVar data validation. Our comprehensive evaluation showed that delivers outstanding performance, with MCC scores predominantly 0.6 0.74. We observed low disordered datasets related CFTR ABC protein. However, superior was shown when benchmarked against quality CFTR2 database. results emphasizes AlphaMissense’s potential pinpointing functional hot spots, its likely surpassing benchmarks calculated from ProteinGym datasets.

Язык: Английский

---

Machine learning-guided co-optimization of fitness and diversity facilitates combinatorial library design in enzyme engineering

Nature Communications, Год журнала: 2024, Номер 15(1)

Опубликована: Июль 29, 2024

Abstract The effective design of combinatorial libraries to balance fitness and diversity facilitates the engineering useful enzyme functions, particularly those that are poorly characterized or unknown in biology. We introduce MODIFY, a machine learning (ML) algorithm learns from natural protein sequences infer evolutionarily plausible mutations predict fitness. MODIFY co-optimizes predicted sequence starting libraries, prioritizing high-fitness variants while ensuring broad coverage. In silico evaluation shows outperforms state-of-the-art unsupervised methods zero-shot prediction enables ML-guided directed evolution with enhanced efficiency. Using we engineer generalist biocatalysts derived thermostable cytochrome c achieve enantioselective C-B C-Si bond formation via new-to-nature carbene transfer mechanism, leading six away previously developed enzymes exhibiting superior comparable activities. These results demonstrate MODIFY’s potential solving challenging problems beyond reach classic evolution.

Язык: Английский

---

Ensembl 2025

Nucleic Acids Research, Год журнала: 2024, Номер 53(D1), С. D948 - D957

Опубликована: Дек. 4, 2024

Ensembl (www.ensembl.org) is an open platform integrating publicly available genomics data across the tree of life with a focus on eukaryotic species related to human health, agriculture and biodiversity. This year has seen continued expansion in number represented, >4800 >31 300 prokaryotic genomes available. The new site, currently beta, develop, holding >2700 genome assemblies. site provides genome, gene, transcript, homology variation views, will replace current Rapid Release site; this represents key step towards provision single integrated site. Additional activities have included developing improved regulatory annotation for human, mouse agricultural species, expanding Variant Effect Predictor tool. To learn more about Ensembl, help documentation are along extensive training program that can be accessed via our pages.

Язык: Английский

---

Human coronavirus HKU1 recognition of the TMPRSS2 host receptor

Cell, Год журнала: 2024, Номер 187(16), С. 4231 - 4245.e13

Опубликована: Июль 3, 2024

The human coronavirus HKU1 spike (S) glycoprotein engages host cell surface sialoglycans and transmembrane protease serine 2 (TMPRSS2) to initiate infection. molecular basis of binding TMPRSS2 determinants receptor tropism remain elusive. We designed an active construct enabling high-yield recombinant production in cells this key therapeutic target. determined a cryo-electron microscopy structure the RBD bound TMPRSS2, providing blueprint interactions supporting viral entry explaining specificity for among orthologous proteases. identified orthologs from five mammalian orders promoting S-mediated into along with residues governing usage. Our data show that motif is site vulnerability neutralizing antibodies suggest uses S conformational masking glycan shielding balance immune evasion engagement.

Язык: Английский

---

Site-saturation mutagenesis of 500 human protein domains

Nature, Год журнала: 2025, Номер unknown

Опубликована: Янв. 8, 2025

Abstract Missense variants that change the amino acid sequences of proteins cause one-third human genetic diseases 1 . Tens millions missense exist in current population, and vast majority these have unknown functional consequences. Here we present a large-scale experimental analysis across many different proteins. Using DNA synthesis cellular selection experiments quantify effect more than 500,000 on abundance 500 protein domains. This dataset reveals 60% pathogenic reduce stability. The contribution stability to fitness varies is particularly important recessive disorders. We combine measurements with language models annotate sites Mutational effects are largely conserved homologous domains, enabling accurate prediction entire families using energy models. Our data demonstrate feasibility assaying at scale provides large consistent reference for clinical variant interpretation training benchmarking computational methods.

Язык: Английский

---

Mechanisms of NLRP3 activation and inhibition elucidated by functional analysis of disease-associated variants

Nature Immunology, Год журнала: 2025, Номер unknown

Опубликована: Фев. 10, 2025

The NLRP3 inflammasome is a multiprotein complex that mediates caspase-1 activation and the release of proinflammatory cytokines, including interleukin (IL)-1β IL-18. Gain-of-function variants in gene encoding (also called cryopyrin) lead to constitutive excessive IL-1β production cryopyrin-associated periodic syndromes (CAPS). Here we present functional screening automated analysis 534 from international INFEVERS registry ClinVar database. This resource captures effect on ASC speck formation spontaneously, at low temperature, after stimulation with specific inhibitor MCC950. Most notably, our facilitated updated classification INFEVERS. Structural suggested multiple mechanisms by which CAPS activate NLRP3, enhanced ATP binding, stabilizing active conformation, destabilizing inactive promoting oligomerization pyrin domain. Furthermore, identified pathogenic can hypersensitize response nigericin cold temperature exposure. We also found most CAPS-related be inhibited MCC950; however, changes proline affecting helices near binding site are resistant MCC950, as domain, likely trigger directly domain ASC. Our findings could help stratify population for clinical trials methodologies implemented molecules different mechanism laboratories worldwide interested adding new functionally validated resource. Overall, study provides improved diagnosis patients CAPS, mechanistic insight into stratification future application targeted therapeutics.

Язык: Английский

---

Artificial intelligence in clinical genetics

European Journal of Human Genetics, Год журнала: 2025, Номер unknown

Опубликована: Янв. 13, 2025

Abstract Artificial intelligence (AI) has been growing more powerful and accessible, will increasingly impact many areas, including virtually all aspects of medicine biomedical research. This review focuses on previous, current, especially emerging applications AI in clinical genetics. Topics covered include a brief explanation different general categories AI, machine learning, deep generative AI. After introductory explanations examples, the discusses genetics three main categories: diagnostics; management therapeutics; support. The concludes with short, medium, long-term predictions about ways that may affect field Overall, while precise speed at which continue to change is unclear, as are overall ramifications for patients, families, clinicians, researchers, others, it likely result dramatic evolution It be important those involved prepare accordingly order minimize risks maximize benefits related use field.

Язык: Английский

---

Predictomes, a classifier-curated database of AlphaFold-modeled protein-protein interactions

Molecular Cell, Год журнала: 2025, Номер unknown

Опубликована: Фев. 1, 2025

Protein-protein interactions (PPIs) are ubiquitous in biology, yet a comprehensive structural characterization of the PPIs underlying cellular processes is lacking. AlphaFold-Multimer (AF-M) has potential to fill this knowledge gap, but standard AF-M confidence metrics do not reliably separate relevant from an abundance false positive predictions. To address limitation, we used machine learning on curated datasets train structure prediction and omics-informed classifier (SPOC) that effectively separates true predictions PPIs, including proteome-wide screens. We applied SPOC all-by-all matrix nearly 300 human genome maintenance proteins, generating ∼40,000 can be viewed at predictomes.org, where users also score their own with SPOC. High-confidence discovered using our approach enable hypothesis generation maintenance. Our results provide framework for interpreting large-scale screens help lay foundation interactome.

Язык: Английский

---

Structure-based self-supervised learning enables ultrafast protein stability prediction upon mutation

The Innovation, Год журнала: 2025, Номер 6(1), С. 100750 - 100750

Опубликована: Янв. 1, 2025

Predicting free energy changes (ΔΔG) is essential for enhancing our understanding of protein evolution and plays a pivotal role in engineering pharmaceutical development. While traditional methods offer valuable insights, they are often constrained by computational speed reliance on biased training datasets. These constraints become particularly evident when aiming accurate ΔΔG predictions across diverse array sequences. Herein, we introduce Pythia, self-supervised graph neural network specifically designed zero-shot predictions. Our comparative benchmarks demonstrate that Pythia outperforms other pretraining models force field-based approaches while also exhibiting competitive performance with fully supervised models. Notably, shows strong correlations achieves remarkable increase up to 105-fold. We further validated Pythia's predicting the thermostabilizing mutations limonene epoxide hydrolase, leading higher experimental success rates. This exceptional efficiency has enabled us explore 26 million high-quality structures, marking significant advancement ability navigate sequence space enhance relationships between genotype phenotype. In addition, established web server at https://pythia.wulab.xyz allow users easily perform such

Язык: Английский

---