Structure prediction of protein-ligand complexes from sequence information with Umol.

Protein-ligand docking is an established tool in drug discovery and development to narrow down potential therapeutics for experimental testing. However, a high-quality protein structure is required and often the protein is treated as fully or partially rigid. Here we develop an AI system that can predict the fully flexible all-atom structure of protein-ligand complexes directly from sequence information. We find that classical docking methods are still superior, but depend upon having crystal structures of the target protein. In addition to predicting flexible all-atom structures, predicted confidence metrics (plDDT) can be used to select accurate predictions as well as to distinguish between strong and weak binders. The advances presented here suggest that the goal of AI-based drug discovery is one step closer, but there is still a way to go to grasp the complexity of protein-ligand interactions fully. Umol is available at: https://github.com/patrickbryant1/Umol .

Consensus Paper: Cerebellum and Ageing.

Given the key roles of the cerebellum in motor, cognitive, and affective operations and given the decline of brain functions with aging, cerebellar circuitry is attracting the attention of the scientific community. The cerebellum plays a key role in timing aspects of both motor and cognitive operations, including for complex tasks such as spatial navigation. Anatomically, the cerebellum is connected with the basal ganglia via disynaptic loops, and it receives inputs from nearly every region in the cerebral cortex. The current leading hypothesis is that the cerebellum builds internal models and facilitates automatic behaviors through multiple interactions with the cerebral cortex, basal ganglia and spinal cord. The cerebellum undergoes structural and functional changes with aging, being involved in mobility frailty and related cognitive impairment as observed in the physio-cognitive decline syndrome (PCDS) affecting older, functionally-preserved adults who show slowness and/or weakness. Reductions in cerebellar volume accompany aging and are at least correlated with cognitive decline. There is a strongly negative correlation between cerebellar volume and age in cross-sectional studies, often mirrored by a reduced performance in motor tasks. Still, predictive motor timing scores remain stable over various age groups despite marked cerebellar atrophy. The cerebello-frontal network could play a significant role in processing speed and impaired cerebellar function due to aging might be compensated by increasing frontal activity to optimize processing speed in the elderly. For cognitive operations, decreased functional connectivity of the default mode network (DMN) is correlated with lower performances. Neuroimaging studies highlight that the cerebellum might be involved in the cognitive decline occurring in Alzheimer's disease (AD), independently of contributions of the cerebral cortex. Grey matter volume loss in AD is distinct from that seen in normal aging, occurring initially in cerebellar posterior lobe regions, and is associated with neuronal, synaptic and beta-amyloid neuropathology. Regarding depression, structural imaging studies have identified a relationship between depressive symptoms and cerebellar gray matter volume. In particular, major depressive disorder (MDD) and higher depressive symptom burden are associated with smaller gray matter volumes in the total cerebellum as well as the posterior cerebellum, vermis, and posterior Crus I. From the genetic/epigenetic standpoint, prominent DNA methylation changes in the cerebellum with aging are both in the form of hypo- and hyper-methylation, and the presumably increased/decreased expression of certain genes might impact on motor coordination. Training influences motor skills and lifelong practice might contribute to structural maintenance of the cerebellum in old age, reducing loss of grey matter volume and therefore contributing to the maintenance of cerebellar reserve. Non-invasive cerebellar stimulation techniques are increasingly being applied to enhance cerebellar functions related to motor, cognitive, and affective operations. They might enhance cerebellar reserve in the elderly. In conclusion, macroscopic and microscopic changes occur in the cerebellum during the lifespan, with changes in structural and functional connectivity with both the cerebral cortex and basal ganglia. With the aging of the population and the impact of aging on quality of life, the panel of experts considers that there is a huge need to clarify how the effects of aging on the cerebellar circuitry modify specific motor, cognitive, and affective operations both in normal subjects and in brain disorders such as AD or MDD, with the goal of preventing symptoms or improving the motor, cognitive, and affective symptoms.

Peptide binder design with inverse folding and protein structure prediction.

The computational design of peptide binders towards a specific protein interface can aid diagnostic and therapeutic efforts. Here, we design peptide binders by combining the known structural space searched with Foldseek, the protein design method ESM-IF1, and AlphaFold2 (AF) in a joint framework. Foldseek generates backbone seeds for a modified version of ESM-IF1 adapted to protein complexes. The resulting sequences are evaluated with AF using an MSA representation for the receptor structure and a single sequence for the binder. We show that AF can accurately evaluate protein binders and that our bind score can select these (ROC AUC = 0.96 for the heterodimeric case). We find that designs created from seeds with more contacts per residue are more successful and tend to be short. There is a relationship between the sequence recovery in interface positions and the plDDT of the designs, where designs with ≥80% recovery have an average plDDT of 84 compared to 55 at 0%. Designed sequences have 60% higher median plDDT values towards intended receptors than non-intended ones. Successful binders (predicted interface RMSD ≤ 2 Å) are designed towards 185 (6.5%) heteromeric and 42 (3.6%) homomeric protein interfaces with ESM-IF1 compared with 18 (1.5%) using ProteinMPNN from 100 samples.

A Swedish Familial Genome-Wide Haplotype Analysis Identified Five Novel Breast Cancer Susceptibility Loci on 9p24.3, 11q22.3, 15q11.2, 16q24.1 and Xq21.31.

Most breast cancer heritability is unexplained. We hypothesized that analysis of unrelated familial cases in a GWAS context could enable the identification of novel susceptibility loci. In order to examine the association of a haplotype with breast cancer risk, we performed a genome-wide haplotype association study using a sliding window analysis of window sizes 1-25 SNPs in 650 familial invasive breast cancer cases and 5021 controls. We identified five novel risk loci on 9p24.3 (OR 3.4; p 4.9 × 10-11), 11q22.3 (OR 2.4; p 5.2 × 10-9), 15q11.2 (OR 3.6; p 2.3 × 10-8), 16q24.1 (OR 3; p 3 × 10-8) and Xq21.31 (OR 3.3; p 1.7 × 10-8) and confirmed three well-known loci on 10q25.13, 11q13.3, and 16q12.1. In total, 1593 significant risk haplotypes and 39 risk SNPs were distributed on the eight loci. In comparison with unselected breast cancer cases from a previous study, the OR was increased in the familial analysis in all eight loci. Analyzing familial cancer cases and controls enabled the identification of novel breast cancer susceptibility loci.

Deep learning for protein complex structure prediction.

Recent developments in the structure prediction of protein complexes have resulted in accuracies rivalling experimental methods in many cases. The high accuracy is mainly observed in dimeric complexes and other problems such as protein disorder and predicting the structure of host-pathogen interactions remain. This review highlights the foundation for current accurate structure prediction of protein complexes and possible ways to address the remaining limitations.

Towards a structurally resolved human protein interaction network.

Cellular functions are governed by molecular machines that assemble through protein-protein interactions. Their atomic details are critical to studying their molecular mechanisms. However, fewer than 5% of hundreds of thousands of human protein interactions have been structurally characterized. Here we test the potential and limitations of recent progress in deep-learning methods using AlphaFold2 to predict structures for 65,484 human protein interactions. We show that experiments can orthogonally confirm higher-confidence models. We identify 3,137 high-confidence models, of which 1,371 have no homology to a known structure. We identify interface residues harboring disease mutations, suggesting potential mechanisms for pathogenic variants. Groups of interface phosphorylation sites show patterns of co-regulation across conditions, suggestive of coordinated tuning of multiple protein interactions as signaling responses. Finally, we provide examples of how the predicted binary complexes can be used to build larger assemblies helping to expand our understanding of human cell biology.

A custom hepatitis A virus assay for whole-genome sequencing.

Since 2016, the United States has experienced a resurgence in the number of hepatitis A virus (HAV) cases and outbreaks. These outbreaks have been sustained by person-to-person transmission with cases occurring predominantly in high-risk populations including intravenous drug users, individuals experiencing homelessness, and men who have sex with men. To investigate HAV transmission, a molecular-surveillance system consisting of real-time RT-PCR (rRT-PCR) for detection, and a conventional RT-PCR assay for genotyping of HAV, was established in New York State (NYS) in 2019. Since then, a total of 271 HAV-positive serum samples collected from cases across NYS between 2019 and 2021 were identified by rRT-PCR. To rapidly and efficiently generate HAV whole-genome sequences, a custom AmpliSeq™ panel was designed in collaboration with Thermo Fisher. To streamline the process, sample preparation was performed on an Ion Chef and sequencing on an Ion S5XL. Of the 271 HAV-positive samples, the whole-genome sequencing (WGS) assay successfully generated 134 near-complete, high-quality HAV sequences. Phylogenetic analysis of the VP1-2A region identified 216 IB, 48 IA, and 2 IIIA genotypes, while 5 were unable to be typed due to poor sequence in this key region. The HAV whole-genome sequencing approach provided a more efficient and streamlined approach for genotyping HAV compared to previous methods and resulted in phylogenetic trees with enhanced resolution compared to the HAV VP1-2A region alone.

Exome sequencing in a Swedish family with PMS2 mutation with varying penetrance of colorectal cancer: investigating the presence of genetic risk modifiers in colorectal cancer risk.

OBJECTIVE: Lynch syndrome is caused by germline mutations in the mismatch repair (MMR) genes, such as the PMS2 gene, and is characterised by a familial accumulation of colorectal cancer. The penetrance of cancer in PMS2 carriers is still not fully elucidated as a colorectal cancer risk has been shown to vary between PMS2 carriers, suggesting the presence of risk modifiers. METHODS: Whole exome sequencing was performed in a Swedish family carrying a PMS2 missense mutation [c.2113G>A, p.(Glu705Lys)]. Thirteen genetic sequence variants were further selected and analysed in a case-control study (724 cases and 711 controls). RESULTS: The most interesting variant was an 18 bp deletion in gene BAG1. BAG1 has been linked to colorectal tumour progression with poor prognosis and is thought to promote colorectal tumour cell survival through increased NF-κB activity. CONCLUSIONS: We conclude the genetic architecture behind the incomplete penetrance of PMS2 is complicated and must be assessed in a genome wide manner using large families and multifactorial analysis.

A structural biology community assessment of AlphaFold2 applications.

Most proteins fold into 3D structures that determine how they function and orchestrate the biological processes of the cell. Recent developments in computational methods for protein structure predictions have reached the accuracy of experimentally determined models. Although this has been independently verified, the implementation of these methods across structural-biology applications remains to be tested. Here, we evaluate the use of AlphaFold2 (AF2) predictions in the study of characteristic structural elements; the impact of missense variants; function and ligand binding site predictions; modeling of interactions; and modeling of experimental structural data. For 11 proteomes, an average of 25% additional residues can be confidently modeled when compared with homology modeling, identifying structural features rarely seen in the Protein Data Bank. AF2-based predictions of protein disorder and complexes surpass dedicated tools, and AF2 models can be used across diverse applications equally well compared with experimentally determined structures, when the confidence metrics are critically considered. In summary, we find that these advances are likely to have a transformative impact in structural biology and broader life-science research.

Predicting the structure of large protein complexes using AlphaFold and Monte Carlo tree search.

AlphaFold can predict the structure of single- and multiple-chain proteins with very high accuracy. However, the accuracy decreases with the number of chains, and the available GPU memory limits the size of protein complexes which can be predicted. Here we show that one can predict the structure of large complexes starting from predictions of subcomponents. We assemble 91 out of 175 complexes with 10-30 chains from predicted subcomponents using Monte Carlo tree search, with a median TM-score of 0.51. There are 30 highly accurate complexes (TM-score ≥0.8, 33% of complete assemblies). We create a scoring function, mpDockQ, that can distinguish if assemblies are complete and predict their accuracy. We find that complexes containing symmetry are accurately assembled, while asymmetrical complexes remain challenging. The method is freely available and accesible as a Colab notebook https://colab.research.google.com/github/patrickbryant1/MoLPC/blob/master/MoLPC.ipynb .

Evaluation of measles IgM antibody detection assays during the 2018-2019 outbreak in New York State.

Although measles was eliminated in the United States in 2000, a severe outbreak occurred between October 2018 and September 2019. New York was especially hard hit. Serology played an integral role in determining immune status (IgG) and identifying, along with molecular analyses, acute measles infections (IgM). Although an indirect immunofluorescence assay (IFA) was historically used by the New York State Department of Health for measles IgM detection, a higher throughput assay was needed to address the increased specimen numbers. Four commercial enzyme-linked immunosorbent assays (ELISAs) were evaluated for sensitivity and specificity in detecting measles IgM. Two ELISA formats were compared, indirect ELISA and IgM antibody capture. Both formats had comparable specificity as determined by cross-reactivity to non-measles specimens. Overall, the sensitivity of the capture ELISAs was greater than the indirect ELISAs and comparable to the indirect immunofluorescence assay with benefits regarding capacity, cost, and turnaround time.

Improved prediction of protein-protein interactions using AlphaFold2.

Predicting the structure of interacting protein chains is a fundamental step towards understanding protein function. Unfortunately, no computational method can produce accurate structures of protein complexes. AlphaFold2, has shown unprecedented levels of accuracy in modelling single chain protein structures. Here, we apply AlphaFold2 for the prediction of heterodimeric protein complexes. We find that the AlphaFold2 protocol together with optimised multiple sequence alignments, generate models with acceptable quality (DockQ ≥ 0.23) for 63% of the dimers. From the predicted interfaces we create a simple function to predict the DockQ score which distinguishes acceptable from incorrect models as well as interacting from non-interacting proteins with state-of-art accuracy. We find that, using the predicted DockQ scores, we can identify 51% of all interacting pairs at 1% FPR.

A Swedish Genome-Wide Haplotype Association Analysis Identifies a Novel Breast Cancer Susceptibility Locus in 8p21.2 and Characterizes Three Loci on Chromosomes 10, 11 and 16.

(1) Background: The heritability of breast cancer is partly explained but much of the genetic contribution remains to be identified. Haplotypes are often used as markers of ethnicity as they are preserved through generations. We have previously demonstrated that haplotype analysis, in addition to standard SNP association studies, could give novel and more detailed information on genetic cancer susceptibility. (2) Methods: In order to examine the association of a SNP or a haplotype to breast cancer risk, we performed a genome wide haplotype association study, using sliding window analysis of window sizes 1−25 and 50 SNPs, in 3200 Swedish breast cancer cases and 5021 controls. (3) Results: We identified a novel breast cancer susceptibility locus in 8p21.1 (OR 2.08; p 3.92 × 10−8), confirmed three known loci in 10q26.13, 11q13.3, 16q12.1-2 and further identified novel subloci within these three loci. Altogether 76 risk SNPs, 3302 risk haplotypes of window size 2−25 and 113 risk haplotypes of window size 50 at p < 5 × 10−8 on chromosomes 8, 10, 11 and 16 were identified. In the known loci haplotype analysis reached an OR of 1.48 in overall breast cancer and in familial cases OR 1.68. (4) Conclusions: Analyzing haplotypes, rather than single variants, could detect novel susceptibility loci even in small study populations but the method requires a fairly homogenous study population.

Vaccine Strain and Wild-Type Clades of Varicella-Zoster Virus in Central Nervous System and Non-CNS Disease, New York State, 2004-2019.

Since the introduction of the varicella-zoster virus (VZV) vaccine in the United States in 1995, there has been a dramatic decrease in both the number and severity of varicella cases. However, VZV surveillance data and information on the VZV clade distribution in central nervous system (CNS) disease and non-CNS disease in New York State is not available. To investigate this, cerebrospinal fluid (CSF) samples from patients with encephalitis or meningitis and non-CSF samples from patients with non-CNS disease manifestations consistent with VZV, collected from 2004 to 2019, were tested with molecular VZV assays. A total of 341 CSF and 1,398 non-CSF samples that tested positive by a VZV-specific real-time PCR assay were further characterized as wild-type or vaccine strain by 3 biallelic real-time PCR assays targeting single nucleotide polymorphism (SNP) markers in open reading frame (ORF) 62. Genotyping was then performed on wild-type strains by conventional PCR and sequencing of 500-bp regions in ORFs 21, 22, and 50. Sequence analysis identified clades 1 to 5 in both sample types with a virtually identical clade distribution between CSF and non-CSF samples. In addition, 19 clade 6 and 13 clade 9 samples were detected in non-CSF samples after implementation of an expanded genotyping scheme, including ORF 29, 38, and 67. These clades were not detected in any CSF samples. Finally, a total of 28 vaccine strains were detected, 25 in the non-CSF samples and 3 in the CSF samples. All three cases of vaccine strain with CNS involvement experienced relatively minor symptoms of aseptic meningitis and fully recovered. These results support the evidence that while the VZV vaccine is capable of causing CNS disease, it is still a rare event and symptoms are typically less severe than those caused by wild-type infection.

The relationship between ageing and changes in the human blood and brain methylomes.

Changes in DNA methylation have been found to be strongly correlated with age, enabling the creation of 'epigenetic clocks'. Previously, studies on the relationship between ageing and DNA methylation have assumed a linear relationship. Here, we show that several markers show a non-linear behaviour. In particular, we observe a tendency for saturation with age, especially in the cerebellum. Further, we show that the relationships between significant methylation changes and ageing are different in different tissues. We suggest a straightforward method of assessing all methylation-age relationships and cluster them according to their relative fold change. Our fold change selection outperforms the most common epigenetic clocks in predicting age for the cerebellum, but not for Blood or the Frontal Cortex. Further, we find that the saturation of methylation observed at older ages for the cerebellum explains why epigenetic clocks consistently underestimate the age there. The findings imply that assuming linear correlations might cause biologically important markers to be missed.

Streamlined Whole-Genome Sequencing of Mumps Virus for High-Resolution Outbreak Analysis.

Since 2015, the United States has experienced a resurgence in the number of mumps cases and outbreaks in fully vaccinated populations. These outbreaks have occurred predominantly in close-quarter settings, such as camps, colleges, and detention centers. Phylogenetic analysis of 758 mumps-positive samples from outbreaks across the United States identified 743 (98%) as genotype G based on sequence analysis of the mumps small hydrophobic (SH) gene. Additionally, SH sequences in the genotype G samples showed almost no sequence diversity, with 675 (91%) of them having identical sequences or only one nucleotide difference. This uniformity of circulating genotype and strain created complications for epidemiologic investigations and necessitated the development of a system for rapidly generating mumps whole-genome sequences for more detailed analysis. In this study, we report a novel and streamlined assay for whole-genome sequencing (WGS) of mumps virus genotype G. The WGS procedure successfully generated 318 high-quality WGS sequences on nucleic acid from genotype G-positive respiratory samples collected during several mumps outbreaks in the United States between 2016 and 2019. Sequencing was performed by a rapid and highly sensitive custom Ion AmpliSeq mumps genotype G panel, with sample preparation performed on an Ion Chef and sequencing on an Ion S5. The WGS data generated by the AmpliSeq panel provided enhanced genomic resolution for epidemiological outbreak investigations. Translation and protein sequence analysis also identified several potentially important epitope changes in the circulating mumps genotype G strains compared to the Jeryl-Lynn strain (JL5) used in vaccines in the United States, which could explain the current level of vaccine escapes.

Congenital Measles in a Premature 25-week Gestation Infant.

We describe a premature infant with congenital measles. Laboratory testing confirmed measles in the mother (polymerase chain reaction- and IgM-positive) and congenital measles in the infant (polymerase chain reaction-positive, culture-positive and IgM-positive). The infant never developed a rash, pneumonia, or neurologic complications. This case supports using compatible laboratory findings to diagnose congenital measles in infants without clinical manifestations of measles.

Estimating the impact of mobility patterns on COVID-19 infection rates in 11 European countries.

BACKGROUND: As governments across Europe have issued non-pharmaceutical interventions (NPIs) such as social distancing and school closing, the mobility patterns in these countries have changed. Most states have implemented similar NPIs at similar time points. However, it is likely different countries and populations respond differently to the NPIs and that these differences cause mobility patterns and thereby the epidemic development to change. METHODS: We build a Bayesian model that estimates the number of deaths on a given day dependent on changes in the basic reproductive number, R 0, due to differences in mobility patterns. We utilise mobility data from Google mobility reports using five different categories: retail and recreation, grocery and pharmacy, transit stations, workplace and residential. The importance of each mobility category for predicting changes in R 0 is estimated through the model. FINDINGS: The changes in mobility have a considerable overlap with the introduction of governmental NPIs, highlighting the importance of government action for population behavioural change. The shift in mobility in all categories shows high correlations with the death rates 1 month later. Reduction of movement within the grocery and pharmacy sector is estimated to account for most of the decrease in R 0. INTERPRETATION: Our model predicts 3-week epidemic forecasts, using real-time observations of changes in mobility patterns, which can provide governments with direct feedback on the effects of their NPIs. The model predicts the changes in a majority of the countries accurately but overestimates the impact of NPIs in Sweden and Denmark and underestimates them in France and Belgium. We also note that the exponential nature of all epidemiological models based on the basic reproductive number, R 0 cause small errors to have extensive effects on the predicted outcome.

Decomposing Structural Response Due to Sequence Changes in Protein Domains with Machine Learning.

How protein domain structure changes in response to mutations is not well understood. Some mutations change the structure drastically, while most only result in small changes. To gain an understanding of this, we decompose the relationship between changes in domain sequence and structure using machine learning. We select pairs of evolutionarily related domains with a broad range of evolutionary distances. In contrast to earlier studies, we do not find a strictly linear relationship between sequence and structural changes. We train a random forest regressor that predicts the structural similarity between pairs with an average accuracy of 0.029 lDDT ( local Distance Difference Test) score, and a correlation coefficient of 0.92. Decomposing the feature importance shows that the domain length, or analogously, size is the most important feature. Our model enables assessing deviations in relative structural response, and thus prediction of evolutionary trajectories, in protein domains across evolution.