GenSLMs: Genome-scale language models reveal SARS-CoV-2 evolutionary dynamics.

We seek to transform how new and emergent variants of pandemic-causing viruses, specifically SARS-CoV-2, are identified and classified. By adapting large language models (LLMs) for genomic data, we build genome-scale language models (GenSLMs) which can learn the evolutionary landscape of SARS-CoV-2 genomes. By pre-training on over 110 million prokaryotic gene sequences and fine-tuning a SARS-CoV-2-specific model on 1.5 million genomes, we show that GenSLMs can accurately and rapidly identify variants of concern. Thus, to our knowledge, GenSLMs represents one of the first whole genome scale foundation models which can generalize to other prediction tasks. We demonstrate scaling of GenSLMs on GPU-based supercomputers and AI-hardware accelerators utilizing 1.63 Zettaflops in training runs with a sustained performance of 121 PFLOPS in mixed precision and peak of 850 PFLOPS. We present initial scientific insights from examining GenSLMs in tracking evolutionary dynamics of SARS-CoV-2, paving the path to realizing this on large biological data.

Neuropeptide-Y Levels in ST-Segment-Elevation Myocardial Infarction: Relationship With Coronary Microvascular Function, Heart Failure, and Mortality.

Background The sympathetic cotransmitter, neuropeptide Y (NPY), is released into the coronary sinus during ST-segment-elevation myocardial infarction and can constrict the coronary microvasculature. We sought to establish whether peripheral venous (PV) NPY levels, which are easy to obtain and measure, are associated with microvascular obstruction, myocardial recovery, and prognosis. Methods and Results NPY levels were measured immediately after primary percutaneous coronary intervention and compared with angiographic and cardiovascular magnetic resonance indexes of microvascular function. Patients were prospectively followed up for 6.4 (interquartile range, 4.1-8.0) years. PV (n=163) and coronary sinus (n=68) NPY levels were significantly correlated (r=0.92; P<0.001) and associated with multiple coronary and imaging parameters of microvascular function and infarct size (such as coronary flow reserve, acute myocardial edema, left ventricular ejection fraction, and late gadolinium enhancement 6 months later). We therefore assessed the prognostic value of PV NPY during follow-up, where 34 patients (20.7%) developed heart failure or died. Kaplan-Meier survival analysis demonstrated that high PV NPY levels (>21.4 pg/mL by binary recursive partitioning) were associated with increased incidence of heart failure and mortality (hazard ratio, 3.49 [95% CI, 1.65-7.4]; P<0.001). This relationship was maintained after adjustment for age, cardiovascular risk factors, and previous myocardial infarction. Conclusions Both PV and coronary sinus NPY levels correlate with microvascular function and infarct size after ST-segment-elevation myocardial infarction. PV NPY levels are associated with the subsequent development of heart failure or mortality and may therefore be a useful prognostic marker. Further research is required to validate these findings.

Intelligent resolution: Integrating Cryo-EM with AI-driven multi-resolution simulations to observe the severe acute respiratory syndrome coronavirus-2 replication-transcription machinery in action.

The severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) replication transcription complex (RTC) is a multi-domain protein responsible for replicating and transcribing the viral mRNA inside a human cell. Attacking RTC function with pharmaceutical compounds is a pathway to treating COVID-19. Conventional tools, e.g., cryo-electron microscopy and all-atom molecular dynamics (AAMD), do not provide sufficiently high resolution or timescale to capture important dynamics of this molecular machine. Consequently, we develop an innovative workflow that bridges the gap between these resolutions, using mesoscale fluctuating finite element analysis (FFEA) continuum simulations and a hierarchy of AI-methods that continually learn and infer features for maintaining consistency between AAMD and FFEA simulations. We leverage a multi-site distributed workflow manager to orchestrate AI, FFEA, and AAMD jobs, providing optimal resource utilization across HPC centers. Our study provides unprecedented access to study the SARS-CoV-2 RTC machinery, while providing general capability for AI-enabled multi-resolution simulations at scale.

ST2 in patients with severe aortic stenosis and heart failure.

BACKGROUND: ST2 is a circulating biomarker that is well established for predicting outcome in heart failure (HF). This is the first study to look at ST2 concentrations in optimally treated patients with stable but significant left ventricular systolic dysfunction (LVSD) compared to patients with severe aortic stenosis (AS). METHODS: Two cohorts were retrospectively studied: 94 patients undergoing transcatheter aortic valve implantation for severe AS (63 with normal ejection fraction [EF] and 31 with reduced EF), and 50 patients with severe LVSD from non-valvular causes. ST2 pre-procedural samples were taken, and repeated again at 3 and 6 months. Patients were followed-up for 2 years. Data was analyzed using SPSS software. RESULTS: Baseline concentrations of soluble ST2 did not differ significantly between the HF group and AS group with normal EF (EF ≥ 50%). However, in the AS group with a low EF (EF < 50%) ST2 concentrations were significantly higher that the HF group (p = 0.009). New York Heart Association class IV HF, baseline N-terminal pro-B-type natriuretic peptide and gender were all independent predictors of soluble ST2 (sST2) baseline concentrations. CONCLUSIONS: Raised ST2 concentrations in the context of severe AS may be a marker for subclinical or clinical left ventricular dysfunction. More research is required to assess its use for assessment of prognosis and response to treatment.

Safety of Rotational Atherectomy Using the Radial Access in Patients With Severe Aortic Stenosis.

Despite frequent percutaneous coronary intervention (PCI) in calcified vessels of older patients, rotational atherectomy (RA) has not been endorsed in patients with severe aortic stenosis (AS) due to safety concerns and lack of data. We explored periprocedural safety and mortality in severe AS patients undergoing RA. Prospective anonymized clinical, echocardiographic, procedural and outcome data of patients undergoing RA PCI between January 2012 and July 2018 were retrospectively extracted from the institutional coronary database. Patients with severe AS undergoing RA PCI were 1:1 propensity matched with patients undergoing RA PCI in the absence of AS. Outcomes of interest were RA related periprocedural complications, 30-day and 1-year mortality. A prespecified subgroup analysis examined the influence of transcatheter aortic valve replacement on mortality following RA PCI. A total of 544 patients underwent RA PCI; 478 without AS and 66 with AS. Propensity matching yielded 35 matched pairs with improved balance in covariates of interest and no significant differences in baseline characteristics postmatching. In the matched cohort (n = 70) slow flow/no-reflow, coronary dissection, perforation, and hemodynamic instability were rare and not significantly different. Survival analyses revealed significantly higher 30-day (Log-Rank p = 0.02) and 1-year mortality (Log rank p = 0.02, HR 5.24 [95% CI 1.13 to 24.28]) in the severe AS group; driven by a fivefold increase in the hazard of death among patients who did not undergo transcatheter aortic valve replacement HR 4.98 [95% CI 1.03 to 24.1]. In conclusion, our study of 70 patients undergoing radial RA PCI suggests that it can be safely performed in patients with severe AS. Long-term outcomes after RA in patients with severe AS are determined by the presence of the valve disease and other co-morbidities.

Prognostic Value of Quantitative Stress Perfusion Cardiac Magnetic Resonance.

OBJECTIVES: This study sought to evaluate the prognostic usefulness of visual and quantitative perfusion cardiac magnetic resonance (CMR) ischemic burden in an unselected group of patients and to assess the validity of consensus-based ischemic burden thresholds extrapolated from nuclear studies. BACKGROUND: There are limited data on the prognostic value of assessing myocardial ischemic burden by CMR, and there are none using quantitative perfusion analysis. METHODS: Patients with suspected coronary artery disease referred for adenosine-stress perfusion CMR were included (n = 395; 70% male; age 58 ± 13 years). The primary endpoint was a composite of cardiovascular death, nonfatal myocardial infarction, aborted sudden death, and revascularization after 90 days. Perfusion scans were assessed visually and with quantitative analysis. Cross-validated Cox regression analysis and net reclassification improvement were used to assess the incremental prognostic value of visual or quantitative perfusion analysis over a baseline clinical model, initially as continuous covariates, then using accepted thresholds of ≥2 segments or ≥10% myocardium. RESULTS: After a median 460 days (interquartile range: 190 to 869 days) follow-up, 52 patients reached the primary endpoint. At 2 years, the addition of ischemic burden was found to increase prognostic value over a baseline model of age, sex, and late gadolinium enhancement (baseline model area under the curve [AUC]: 0.75; visual AUC: 0.84; quantitative AUC: 0.85). Dichotomized quantitative ischemic burden performed better than visual assessment (net reclassification improvement 0.043 vs. 0.003 against baseline model). CONCLUSIONS: This study was the first to address the prognostic benefit of quantitative analysis of perfusion CMR and to support the use of consensus-based ischemic burden thresholds by perfusion CMR for prognostic evaluation of patients with suspected coronary artery disease. Quantitative analysis provided incremental prognostic value to visual assessment and established risk factors, potentially representing an important step forward in the translation of quantitative CMR perfusion analysis to the clinical setting.