Large language models for science and medicine.

Large language models (LLMs) are a type of machine learning model that learn statistical patterns over text, such as predicting the next words in a sequence of text. Both general purpose and task-specific LLMs have demonstrated potential across diverse applications. Science and medicine have many data types that are highly suitable for LLMs, such as scientific texts (publications, patents and textbooks), electronic medical records, large databases of DNA and protein sequences and chemical compounds. Carefully validated systems that can understand and reason across all these modalities may maximize benefits. Despite the inevitable limitations and caveats of any new technology and some uncertainties specific to LLMs, LLMs have the potential to be transformative in science and medicine.

Human coronavirus HKU1 recognition of the TMPRSS2 host receptor.

The human coronavirus HKU1 spike (S) glycoprotein engages host cell surface sialoglycans and transmembrane protease serine 2 (TMPRSS2) to initiate infection. The molecular basis of HKU1 binding to TMPRSS2 and determinants of host receptor tropism remain elusive. Here, we designed an active human TMPRSS2 construct enabling high-yield recombinant production in human cells of this key therapeutic target. We determined a cryo-electron microscopy structure of the HKU1 RBD bound to human TMPRSS2 providing a blueprint of the interactions supporting viral entry and explaining the specificity for TMPRSS2 among human type 2 transmembrane serine proteases. We found that human, rat, hamster and camel TMPRSS2 promote HKU1 S-mediated entry into cells and identified key residues governing host receptor usage. Our data show that serum antibodies targeting the HKU1 RBD TMPRSS2 binding-site are key for neutralization and that HKU1 uses conformational masking and glycan shielding to balance immune evasion and receptor engagement.

Epidemiology and determinants of reemerging bacterial sexually transmitted infections (STIs) and emerging STIs in Europe.

In this scoping review, we offer a comprehensive understanding of the current and recent epidemiology, challenges, and emerging issues related to bacterial sexually transmitted infections (STIs) in the WHO European Region. We endeavour in collating data from both EU/EEA and non- EU/EEA countries, thereby giving a complete picture of the region which highlights the higher notification rates in Northern and Western countries than other regions, likely due to differences in testing, access to testing, and surveillance capacity. We provide an up-to-date review on the current knowledge of determinants and persistent inequities in key populations as well as the use of molecular epidemiology for identifying transmission networks in gonorrhoea and syphilis, and detecting chlamydia mutations that evade molecular diagnosis. Finally, we explore the emerging STIs in the region and the evolving transmission routes of food and waterborne diseases into sexual transmission. Our findings call for harmonized STI surveillance systems, proactive strategies, and policies to address social factors, and staying vigilant for emerging STIs.

A framework for future national pediatric pandemic respiratory disease severity triage: The HHS pediatric COVID-19 data challenge.

Introduction: With persistent incidence, incomplete vaccination rates, confounding respiratory illnesses, and few therapeutic interventions available, COVID-19 continues to be a burden on the pediatric population. During a surge, it is difficult for hospitals to direct limited healthcare resources effectively. While the overwhelming majority of pediatric infections are mild, there have been life-threatening exceptions that illuminated the need to proactively identify pediatric patients at risk of severe COVID-19 and other respiratory infectious diseases. However, a nationwide capability for developing validated computational tools to identify pediatric patients at risk using real-world data does not exist. Methods: HHS ASPR BARDA sought, through the power of competition in a challenge, to create computational models to address two clinically important questions using the National COVID Cohort Collaborative: (1) Of pediatric patients who test positive for COVID-19 in an outpatient setting, who are at risk for hospitalization? (2) Of pediatric patients who test positive for COVID-19 and are hospitalized, who are at risk for needing mechanical ventilation or cardiovascular interventions? Results: This challenge was the first, multi-agency, coordinated computational challenge carried out by the federal government as a response to a public health emergency. Fifty-five computational models were evaluated across both tasks and two winners and three honorable mentions were selected. Conclusion: This challenge serves as a framework for how the government, research communities, and large data repositories can be brought together to source solutions when resources are strapped during a pandemic.

Potent broadly neutralizing antibody VIR-3434 controls hepatitis B and D virus infection and reduces HBsAg in humanized mice.

BACKGROUND & AIMS: Chronic hepatitis B is a global public health problem, and coinfection with hepatitis delta virus (HDV) worsens disease outcome. Here, we describe a hepatitis B virus (HBV) surface antigen (HBsAg)-targeting monoclonal antibody (mAb) with the potential to treat chronic hepatitis B and chronic hepatitis D. METHODS: HBsAg-specific mAbs were isolated from memory B cells of HBV vaccinated individuals. In vitro neutralization was determined against HBV and HDV enveloped with HBsAg representing eight HBV genotypes. Human liver-chimeric mice were treated twice weekly with a candidate mAb starting 3 weeks post HBV inoculation (spreading phase) or during stable HBV or HBV/HDV coinfection (chronic phase). RESULTS: From a panel of human anti-HBs mAbs, VIR-3434 was selected and engineered for pre-clinical development. VIR-3434 targets a conserved, conformational epitope within the antigenic loop of HBsAg and neutralized HBV and HDV infection with higher potency than hepatitis B immunoglobulins in vitro. Neutralization was pan-genotypic against strains representative of HBV genotypes A-H. In the spreading phase of HBV infection in human liver-chimeric mice, a parental mAb of VIR-3434 (HBC34) prevented HBV dissemination and the increase in intrahepatic HBV RNA and covalently closed circular DNA. In the chronic phase of HBV infection or co-infection with HDV, HBC34 treatment decreased circulating HBsAg by >1 log and HDV RNA by >2 logs. CONCLUSIONS: The potently neutralizing anti-HBs mAb VIR-3434 reduces circulating HBsAg and HBV/HDV viremia in human liver-chimeric mice. VIR-3434 is currently in clinical development for treatment of patients with chronic hepatitis B or D. IMPACT AND IMPLICATIONS: Chronic infection with hepatitis B virus and co-infection with hepatitis D virus place approximately 290 million individuals worldwide at risk of severe liver disease and cancer. Available treatments result in low rates of functional cure or require lifelong therapy that does not eliminate the risk of liver disease. We isolated and characterized a potent human antibody that neutralizes hepatitis B and D viruses and reduces infection in a mouse model. This antibody could provide a new treatment for patients with chronic hepatitis B and D.

Natural heteroclitic-like peptides are generated by SARS-CoV-2 mutations.

Humoral immunity is sensitive to evasion by SARS-CoV-2 mutants, but CD8 T cells seem to be more resistant to mutational inactivation. By a systematic analysis of 30 spike variant peptides containing the most relevant VOC and VOI mutations that have accumulated overtime, we show that in vaccinated and convalescent subjects, mutated epitopes can have not only a neutral or inhibitory effect on CD8 T cell recognition but can also enhance or generate de novo CD8 T cell responses. The emergence of these mutated T cell function enhancing epitopes likely reflects an epiphenomenon of SARS-CoV-2 evolution driven by antibody evasion and increased virus transmissibility. In a subset of individuals with weak and narrowly focused CD8 T cell responses selection of these heteroclitic-like epitopes may bear clinical relevance by improving antiviral protection. The functional enhancing effect of these peptides is also worth of consideration for the future development of new generation, more potent COVID-19 vaccines.

A pan-influenza antibody inhibiting neuraminidase via receptor mimicry.

Rapidly evolving influenza A viruses (IAVs) and influenza B viruses (IBVs) are major causes of recurrent lower respiratory tract infections. Current influenza vaccines elicit antibodies predominantly to the highly variable head region of haemagglutinin and their effectiveness is limited by viral drift1 and suboptimal immune responses2. Here we describe a neuraminidase-targeting monoclonal antibody, FNI9, that potently inhibits the enzymatic activity of all group 1 and group 2 IAVs, as well as Victoria/2/87-like, Yamagata/16/88-like and ancestral IBVs. FNI9 broadly neutralizes seasonal IAVs and IBVs, including the immune-evading H3N2 strains bearing an N-glycan at position 245, and shows synergistic activity when combined with anti-haemagglutinin stem-directed antibodies. Structural analysis reveals that D107 in the FNI9 heavy chain complementarity-determinant region 3 mimics the interaction of the sialic acid carboxyl group with the three highly conserved arginine residues (R118, R292 and R371) of the neuraminidase catalytic site. FNI9 demonstrates potent prophylactic activity against lethal IAV and IBV infections in mice. The unprecedented breadth and potency of the FNI9 monoclonal antibody supports its development for the prevention of influenza illness by seasonal and pandemic viruses.

Real-World Effectiveness of Sotrovimab for the Early Treatment of COVID-19 During SARS-CoV-2 Delta and Omicron Waves in the USA.

INTRODUCTION: Sotrovimab, a recombinant human monoclonal antibody (mAb) against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) had US Food and Drug Administration Emergency Use Authorization for the treatment of high-risk outpatients with mild-to-moderate coronavirus disease 2019 (COVID-19) from 26 May 2021 to 5 April 2022. Real-world clinical effectiveness of sotrovimab in reducing the risk of 30-day all-cause hospitalization and/or mortality was evaluated for the period when the prevalence of circulating SARS-CoV-2 variants changed between Delta and Omicron in the USA. METHODS: A retrospective analysis was conducted of de-identified patients diagnosed with COVID-19 between 1 September 2021 to 30 April 2022 in the FAIR Health National Private Insurance Claims database. Patients meeting high-risk criteria were divided into two cohorts: sotrovimab and not treated with a mAb ("no mAb"). All-cause hospitalizations and facility-reported mortality ≤ 30 days of diagnosis ("30-day hospitalization or mortality") were identified. Multivariable and propensity score-matched Poisson and logistic regressions were conducted to estimate the adjusted relative risk (RR) and odds of 30-day hospitalization or mortality in each cohort. RESULTS: Compared with the no mAb cohort (n = 1,514,868), the sotrovimab cohort (n = 15,633) was older and had a higher proportion of patients with high-risk conditions. In the no mAb cohort, 84,307 (5.57%) patients were hospitalized and 8167 (0.54%) deaths were identified, while in the sotrovimab cohort, 418 (2.67%) patients were hospitalized and 13 (0.08%) deaths were identified. After adjusting for potential confounders, the sotrovimab cohort had a 55% lower risk of 30-day hospitalization or mortality (RR 0.45, 95% CI 0.41-0.49) and an 85% lower risk of 30-day mortality (RR 0.15, 95% CI 0.08-0.29). Monthly, from September 2021 to April 2022, the RR reduction for 30-day hospitalization or mortality in the sotrovimab cohort was maintained, ranging from 46% to 71% compared with the no mAb cohort; the RR estimate in April 2022 was uncertain, with wide confidence intervals due to the small sample size. CONCLUSION: Sotrovimab was associated with reduced risk of 30-day all-cause hospitalization and mortality versus no mAb treatment. Clinical effectiveness persisted during Delta and early Omicron variant waves and among all high-risk subgroups assessed.

Maturation of SARS-CoV-2 Spike-specific memory B cells drives resilience to viral escape.

Memory B cells (MBCs) generate rapid antibody responses upon secondary encounter with a pathogen. Here, we investigated the kinetics, avidity, and cross-reactivity of serum antibodies and MBCs in 155 SARS-CoV-2 infected and vaccinated individuals over a 16-month time frame. SARS-CoV-2-specific MBCs and serum antibodies reached steady-state titers with comparable kinetics in infected and vaccinated individuals. Whereas MBCs of infected individuals targeted both prefusion and postfusion Spike (S), most vaccine-elicited MBCs were specific for prefusion S, consistent with the use of prefusion-stabilized S in mRNA vaccines. Furthermore, a large fraction of MBCs recognizing postfusion S cross-reacted with human betacoronaviruses. The avidity of MBC-derived and serum antibodies increased over time resulting in enhanced resilience to viral escape by SARS-CoV-2 variants, including Omicron BA.1 and BA.2 sublineages, albeit only partially for BA.4 and BA.5 sublineages. Overall, the maturation of high-affinity and broadly reactive MBCs provides the basis for effective recall responses to future SARS-CoV-2 variants.

Ensuring Trust in Genomics Research.

Reproducibility, transparency, representation, and privacy underpin the trust on genomics research in general and genome-wide association studies (GWAS) in particular. Concerns about these issues can be mitigated by technologies that address privacy protection, quality control, and verifiability of GWAS. However, many of the existing technological solutions have been developed in isolation and may address one aspect of reproducibility, transparency, representation, and privacy of GWAS while unknowingly impacting other aspects. As a consequence, the current patchwork of technological tools only partially and in an overlapping manner address issues with GWAS, sometimes even creating more problems. This paper addresses the progress in a field that creates technological solutions that augment the acceptance and security of population genetic analyses. The text identifies areas that are falling behind in technical implementation or where there is insufficient research. We make the case that a full understanding of the different GWAS settings, technological tools and new research directions can holistically address the requirements for the acceptance of GWAS.

Severe acute infection and chronic pulmonary disease are risk factors for developing post-COVID-19 conditions.

Post-COVID-19 conditions, also known as "long COVID", has significantly impacted the lives of many individuals, but the risk factors for this condition are poorly understood. In this study, we performed a retrospective EHR analysis of 89,843 individuals at a multi-state health system in the United States with PCR-confirmed COVID-19, including 1,086 patients diagnosed with long COVID and 1,086 matched controls not diagnosed with long COVID. For these two cohorts, we evaluated a wide range of clinical covariates, including laboratory tests, medication orders, phenotypes recorded in the clinical notes, and outcomes. We found that chronic pulmonary disease (CPD) was significantly more common as a pre-existing condition for the long COVID cohort than the control cohort (odds ratio: 1.9, 95% CI: [1.5, 2.6]). Additionally, long-COVID patients were more likely to have a history of migraine (odds ratio: 2.2, 95% CI: [1.6, 3.1]) and fibromyalgia (odds ratio: 2.3, 95% CI: [1.3, 3.8]). During the acute infection phase, the following lab measurements were abnormal in the long COVID cohort: high triglycerides (meanlongCOVID: 278.5 mg/dL vs. meancontrol: 141.4 mg/dL), low HDL cholesterol levels (meanlongCOVID: 38.4 mg/dL vs. meancontrol: 52.5 mg/dL), and high neutrophil-lymphocyte ratio (meanlongCOVID: 10.7 vs. meancontrol: 7.2). The hospitalization rate during the acute infection phase was also higher in the long COVID cohort compared to the control cohort (ratelongCOVID: 5% vs. ratecontrol: 1%). Overall, this study suggests that the severity of acute infection and a history of CPD, migraine, CFS, or fibromyalgia may be risk factors for long COVID symptoms. Our findings motivate clinical studies to evaluate whether suppressing acute disease severity proactively, especially in patients at high risk, can reduce incidence of long COVID.

Maturation of SARS-CoV-2 Spike-specific memory B cells drives resilience to viral escape.

Memory B cells (MBCs) generate rapid antibody responses upon secondary encounter with a pathogen. Here, we investigated the kinetics, avidity and cross-reactivity of serum antibodies and MBCs in 155 SARS-CoV-2 infected and vaccinated individuals over a 16-month timeframe. SARS-CoV-2-specific MBCs and serum antibodies reached steady-state titers with comparable kinetics in infected and vaccinated individuals. Whereas MBCs of infected individuals targeted both pre- and postfusion Spike (S), most vaccine-elicited MBCs were specific for prefusion S, consistent with the use of prefusion-stabilized S in mRNA vaccines. Furthermore, a large fraction of MBCs recognizing postfusion S cross-reacted with human betacoronaviruses. The avidity of MBC-derived and serum antibodies increased over time resulting in enhanced resilience to viral escape by SARS-CoV-2 variants, including Omicron BA.1 and BA.2 sub-lineages, albeit only partially for BA.4 and BA.5 sublineages. Overall, the maturation of high-affinity and broadly-reactive MBCs provides the basis for effective recall responses to future SARS-CoV-2 variants.

Antibody therapy reverses biological signatures of COVID-19 progression.

Understanding who is at risk of progression to severe coronavirus disease 2019 (COVID-19) is key to clinical decision making and effective treatment. We study correlates of disease severity in the COMET-ICE clinical trial that randomized 1:1 to placebo or to sotrovimab, a monoclonal antibody for the treatment of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection (ClinicalTrials.gov04545060). Laboratory parameters identify study participants at greater risk of severe disease, including a high neutrophil-to-lymphocyte ratio (NLR), a negative SARS-CoV-2 serologic test, and whole-blood transcriptome profiles. Sotrovimab treatment is associated with normalization of NLR and the transcriptomic profile and with a decrease of viral RNA in nasopharyngeal samples. Transcriptomics provides the most sensitive detection of participants who would go on to be hospitalized or die. To facilitate timely measurement, we identify a 10-gene signature with similar predictive accuracy. We identify markers of risk for disease progression and demonstrate that normalization of these parameters occurs with antibody treatment of established infection.

Structure, receptor recognition, and antigenicity of the human coronavirus CCoV-HuPn-2018 spike glycoprotein.

The isolation of CCoV-HuPn-2018 from a child respiratory swab indicates that more coronaviruses are spilling over to humans than previously appreciated. We determined the structures of the CCoV-HuPn-2018 spike glycoprotein trimer in two distinct conformational states and showed that its domain 0 recognizes sialosides. We identified that the CCoV-HuPn-2018 spike binds canine, feline, and porcine aminopeptidase N (APN) orthologs, which serve as entry receptors, and determined the structure of the receptor-binding B domain in complex with canine APN. The introduction of an oligosaccharide at position N739 of human APN renders cells susceptible to CCoV-HuPn-2018 spike-mediated entry, suggesting that single-nucleotide polymorphisms might account for viral detection in some individuals. Human polyclonal plasma antibodies elicited by HCoV-229E infection and a porcine coronavirus monoclonal antibody inhibit CCoV-HuPn-2018 spike-mediated entry, underscoring the cross-neutralizing activity among ɑ-coronaviruses. These data pave the way for vaccine and therapeutic development targeting this zoonotic pathogen representing the eighth human-infecting coronavirus.

The Evolution and Biology of SARS-CoV-2 Variants.

Our understanding of the still unfolding severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic would have been extremely limited without the study of the genetics and evolution of this new human coronavirus. Large-scale genome-sequencing efforts have provided close to real-time tracking of the global spread and diversification of SARS-CoV-2 since its entry into the human population in late 2019. These data have underpinned analysis of its origins, epidemiology, and adaptations to the human population: principally immune evasion and increasing transmissibility. SARS-CoV-2, despite being a new human pathogen, was highly capable of human-to-human transmission. During its rapid spread in humans, SARS-CoV-2 has evolved independent new forms, the so-called "variants of concern," that are better optimized for human-to-human transmission. The most important adaptation of the bat coronavirus progenitor of both SARS-CoV-1 and SARS-CoV-2 for human infection (and other mammals) is the use of the angiotensin-converting enzyme 2 (ACE2) receptor. Relaxed structural constraints provide plasticity to SARS-related coronavirus spike protein permitting it to accommodate significant amino acid replacements of antigenic consequence without compromising the ability to bind to ACE2. Although the bulk of research has justifiably concentrated on the viral spike protein as the main determinant of antigenic evolution and changes in transmissibility, there is accumulating evidence for the contribution of other regions of the viral proteome to virus-host interaction. Whereas levels of community transmission of recombinants compromising genetically distinct variants are at present low, when divergent variants cocirculate, recombination between SARS-CoV-2 clades is being detected, increasing the risk that viruses with new properties emerge. Applying computational and machine learning methods to genome sequence data sets to generate experimentally verifiable predictions will serve as an early warning system for novel variant surveillance and will be important in future vaccine planning. Omicron, the latest SARS-CoV-2 variant of concern, has focused attention on step change antigenic events, "shift," as opposed to incremental "drift" changes in antigenicity. Both an increase in transmissibility and antigenic shift in Omicron led to it readily causing infections in the fully vaccinated and/or previously infected. Omicron's virulence, while reduced relative to the variant of concern it replaced, Delta, is very much premised on the past immune exposure of individuals with a clear signal that boosted vaccination protects from severe disease. Currently, SARS-CoV-2 has proven itself to be a dangerous new human respiratory pathogen with an unpredictable evolutionary capacity, leading to a risk of future variants too great not to ensure all regions of the world are screened by viral genome sequencing, protected through available and affordable vaccines, and have non-punitive strategies in place for detecting and responding to novel variants of concern.

Predicting the mutational drivers of future SARS-CoV-2 variants of concern.

SARS-CoV-2 evolution threatens vaccine- and natural infection-derived immunity as well as the efficacy of therapeutic antibodies. To improve public health preparedness, we sought to predict which existing amino acid mutations in SARS-CoV-2 might contribute to future variants of concern. We tested the predictive value of features comprising epidemiology, evolution, immunology, and neural network-based protein sequence modeling, and identified primary biological drivers of SARS-CoV-2 intra-pandemic evolution. We found evidence that ACE2-mediated transmissibility and resistance to population-level host immunity has waxed and waned as a primary driver of SARS-CoV-2 evolution over time. We retroactively identified with high accuracy (area under the receiver operator characteristic curve, AUROC=0.92-0.97) mutations that will spread, at up to four months in advance, across different phases of the pandemic. The behavior of the model was consistent with a plausible causal structure wherein epidemiological covariates combine the effects of diverse and shifting drivers of viral fitness. We applied our model to forecast mutations that will spread in the future and characterize how these mutations affect the binding of therapeutic antibodies. These findings demonstrate that it is possible to forecast the driver mutations that could appear in emerging SARS-CoV-2 variants of concern. We validate this result against Omicron, showing elevated predictive scores for its component mutations prior to emergence, and rapid score increase across daily forecasts during emergence. This modeling approach may be applied to any rapidly evolving pathogens with sufficiently dense genomic surveillance data, such as influenza, and unknown future pandemic viruses.

MP-NeRF: A massively parallel method for accelerating protein structure reconstruction from internal coordinates.

The conversion of proteins between internal and cartesian coordinates is a limiting step in many pipelines, such as molecular dynamics simulations and machine learning models. This conversion is typically carried out by sequential or parallel applications of the Natural extension of Reference Frame (NeRF) algorithm. This work proposes a massively parallel NeRF implementation which, depending on the polymer length, achieves speedups between 400 and 1200× over the previous state-of-the-art. It accomplishes this by dividing the conversion into three main phases: parallel composition of the monomer backbone, assembly of backbone subunits, and parallel elongation of sidechains; and by batching these computations into a minimal number of efficient matrix operations. Special emphasis is placed on reusability and ease of use. We open source the code (available at https://github.com/EleutherAI/mp_nerf) and provide a corresponding python package.