Accurate structure prediction of biomolecular interactions with AlphaFold 3.

The introduction of AlphaFold 21 has spurred a revolution in modelling the structure of proteins and their interactions, enabling a huge range of applications in protein modelling and design2-6. Here we describe our AlphaFold 3 model with a substantially updated diffusion-based architecture that is capable of predicting the joint structure of complexes including proteins, nucleic acids, small molecules, ions and modified residues. The new AlphaFold model demonstrates substantially improved accuracy over many previous specialized tools: far greater accuracy for protein-ligand interactions compared with state-of-the-art docking tools, much higher accuracy for protein-nucleic acid interactions compared with nucleic-acid-specific predictors and substantially higher antibody-antigen prediction accuracy compared with AlphaFold-Multimer v.2.37,8. Together, these results show that high-accuracy modelling across biomolecular space is possible within a single unified deep-learning framework.

Applying and improving AlphaFold at CASP14.

We describe the operation and improvement of AlphaFold, the system that was entered by the team AlphaFold2 to the "human" category in the 14th Critical Assessment of Protein Structure Prediction (CASP14). The AlphaFold system entered in CASP14 is entirely different to the one entered in CASP13. It used a novel end-to-end deep neural network trained to produce protein structures from amino acid sequence, multiple sequence alignments, and homologous proteins. In the assessors' ranking by summed z scores (>2.0), AlphaFold scored 244.0 compared to 90.8 by the next best group. The predictions made by AlphaFold had a median domain GDT_TS of 92.4; this is the first time that this level of average accuracy has been achieved during CASP, especially on the more difficult Free Modeling targets, and represents a significant improvement in the state of the art in protein structure prediction. We reported how AlphaFold was run as a human team during CASP14 and improved such that it now achieves an equivalent level of performance without intervention, opening the door to highly accurate large-scale structure prediction.

Highly accurate protein structure prediction with AlphaFold.

Proteins are essential to life, and understanding their structure can facilitate a mechanistic understanding of their function. Through an enormous experimental effort1-4, the structures of around 100,000 unique proteins have been determined5, but this represents a small fraction of the billions of known protein sequences6,7. Structural coverage is bottlenecked by the months to years of painstaking effort required to determine a single protein structure. Accurate computational approaches are needed to address this gap and to enable large-scale structural bioinformatics. Predicting the three-dimensional structure that a protein will adopt based solely on its amino acid sequence-the structure prediction component of the 'protein folding problem'8-has been an important open research problem for more than 50 years9. Despite recent progress10-14, existing methods fall far short of atomic accuracy, especially when no homologous structure is available. Here we provide the first computational method that can regularly predict protein structures with atomic accuracy even in cases in which no similar structure is known. We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Critical Assessment of protein Structure Prediction (CASP14)15, demonstrating accuracy competitive with experimental structures in a majority of cases and greatly outperforming other methods. Underpinning the latest version of AlphaFold is a novel machine learning approach that incorporates physical and biological knowledge about protein structure, leveraging multi-sequence alignments, into the design of the deep learning algorithm.

Highly accurate protein structure prediction for the human proteome.

Protein structures can provide invaluable information, both for reasoning about biological processes and for enabling interventions such as structure-based drug development or targeted mutagenesis. After decades of effort, 17% of the total residues in human protein sequences are covered by an experimentally determined structure1. Here we markedly expand the structural coverage of the proteome by applying the state-of-the-art machine learning method, AlphaFold2, at a scale that covers almost the entire human proteome (98.5% of human proteins). The resulting dataset covers 58% of residues with a confident prediction, of which a subset (36% of all residues) have very high confidence. We introduce several metrics developed by building on the AlphaFold model and use them to interpret the dataset, identifying strong multi-domain predictions as well as regions that are likely to be disordered. Finally, we provide some case studies to illustrate how high-quality predictions could be used to generate biological hypotheses. We are making our predictions freely available to the community and anticipate that routine large-scale and high-accuracy structure prediction will become an important tool that will allow new questions to be addressed from a structural perspective.

A novel exon 2 I27V VCP variant is associated with dissimilar clinical syndromes.

Mutations in valosin-containing protein (VCP) are associated with a syndromic constellation of inclusion body myositis, Paget's disease of bone and frontotemporal dementia. Here we describe the case reports of two patients with a novel variation (p.I27V) in the VCP gene that was not identified in a healthy control population. One patient presented with a frontotemporal dementia syndrome associated with raised serum alkaline phosphatase and a family history of progressive muscle disease and behavioural decline, while the second patient presented with isolated progressive dysarthria. Together these cases suggest a potential for the same VCP mutation to produce distinct patterns of brain damage, underlining the clinical heterogeneity of VCP-associated disease.

Peripheral delivery of a ROCK inhibitor improves learning and working memory.

Previously, utilizing a series of genome-wide association, brain imaging, and gene expression studies we implicated the KIBRA gene and the RhoA/ROCK pathway in hippocampal-mediated human memory. Here we show that peripheral administration of the ROCK inhibitor hydroxyfasudil improves spatial learning and working memory in the rodent model. This study supports the action of ROCK on learning and memory, suggests the potential value of ROCK inhibition for the promotion of cognition in humans, and highlights the powerful potential of unbiased genome-wide association studies to inform potential novel uses for existing pharmaceuticals.

Identification of somatic chromosomal abnormalities in hypothalamic hamartoma tissue at the GLI3 locus.

Hypothalamic hamartomas (HH) are rare, benign congenital tumors associated with intractable epilepsy. Most cases are sporadic and nonsyndromic. Approximately 5% of HH cases are associated with Pallister-Hall syndrome (PHS), which is caused by haploinsufficiency of GLI3. We have investigated the possibility that HH pathogenesis in sporadic cases is due to a somatic (tumor-only) mutation in GLI3. We isolated genomic DNA from peripheral blood and surgically resected HH tissue in 55 patients with sporadic HH and intractable epilepsy. A genome-wide screen for loss of heterozygosity (LOH) and chromosomal abnormalities was performed with parallel analysis of blood and HH tissue with Affymetrix 10K SNP microarrays. Additionally, resequencing and fine mapping with SNP genotyping were completed for the GLI3 gene with comparisons between peripheral blood and HH tissue pairs. By analyzing chromosomal copy-number data for paired samples on the Affymetrix 10K array, we identified a somatic chromosomal abnormality on chromosome 7p in one HH tissue sample. Resequencing of GLI3 did not identify causative germline mutations but did identify LOH within the GLI3 gene in the HH tissue samples of three patients. Further genotyping of 28 SNPs within and surrounding GLI3 identified five additional patients exhibiting LOH. Together, these data provide evidence that the development of chromosomal abnormalities within GLI3 is associated with the pathogenesis of HH lesions in sporadic, nonsyndromic patients with HH and intractable epilepsy. Chromosomal abnormalities including the GLI3 locus were seen in 8 of 55 (15%) of the resected HH tissue samples. These somatic mutations appear to be highly variable.

Chromosomal abnormality at 6p25.1-25.3 identifies a susceptibility locus for hypothalamic hamartoma associated with epilepsy.

The pathogenesis of hypothalamic hamartoma (HH) associated with epilepsy is unknown. We have identified an individual with HH and refractory epilepsy exhibiting subtle dysmorphic features. High-resolution karyotype identified a duplication of the terminal end of 6p (6p25.1-25.3), confirmed by fluorescent in situ-hybridization (FISH). Copy number analysis with high-density (250K) single nucleotide polymorphism (SNP) genotyping microarrays characterized the abnormality as a series of amplified regions between 1.4 Mb and 10.2 Mb, with a small tandem deletion from 8.8 Mb to 9.7 Mb. There are 38 RefSeq genes within the duplicated regions, and no known coding sequences within the deletion. This unique patient helps identify 6p25.1-25.3 as a possible susceptibility locus for sporadic HH.