Explore the vast range of titles available.

Potts models and variational autoencoders (VAEs) have recently gained popularity as generative protein sequence models (GPSMs) to explore fitness landscapes and predict mutation effects. Despite encouraging results, current model evaluation metrics leave unclear whether GPSMs faithfully reproduce the complex multi-residue mutational patterns observed in natural sequences due to epistasis. Here, we develop a set of sequence statistics to assess the “generative capacity” of three current GPSMs: the pairwise Potts Hamiltonian, the VAE, and the site-independent model. We show that the Potts model’s generative capacity is largest, as the higher-order mutational statistics generated by the model agree with those observed for natural sequences, while the VAE’s lies between the Potts and site-independent models. Importantly, our work provides a new framework for evaluating and interpreting GPSM accuracy which emphasizes the role of higher-order covariation and epistasis, with broader implications for probabilistic sequence models in general. Generative models have become increasingly popular in protein design, yet rigorous metrics that allow the comparison of these models are lacking. Here, the authors propose a set of such metrics and use them to compare three popular models.

Transient receptor potential (TRP) proteins are a large family of cation-selective channels, surpassed in variety only by voltage-gated potassium channels. Detailed molecular mechanisms governing how membrane voltage, ligand binding, or temperature can induce conformational changes promoting the open state in TRP channels are still a matter of debate. Aiming to unveil distinctive structural features common to the transmembrane domains within the TRP family, we performed phylogenetic reconstruction, sequence statistics, and structural analysis over a large set of TRP channel genes. Here, we report an exceptionally conserved set of residues. This fingerprint is composed of twelve residues localized at equivalent three-dimensional positions in TRP channels from the different subtypes. Moreover, these amino acids are arranged in three groups, connected by a set of aromatics located at the core of the transmembrane structure. We hypothesize that differences in the connectivity between these different groups of residues harbor the apparent differences in coupling strategies used by TRP subgroups.

The convergence of advanced models, data, and computational resources are allowing generative artificial intelligence (AI) to revolutionize science and society around text, images, and molecules. This work combines a theoretical generative AI methodology, a practical application of bioinformatics, and an optimized high-throughput computational tool to provide insights and conduct explorations into the protein sequence-structure-function continuum.

Multidrug resistant (MDR) strains of Pseudomonas aeruginosa are a growing threat to humans. P. aeruginosa coordinates a wide variety of behavior through the LasI/LasR quorum sensing pathway. This includes behavior that is virulent to host cells. Acyl homoserine lactones (AHLs) are autoinducers of the LasI/LasR pathway. After synthesis by LasI, a LuxI-type acyl homoserine lactone (AHL) synthase, AHLs are secreted to the extracellular space. They re-enter neighboring bacteria and activate the quorum sensing receptor LasR, the major transcriptional regulator for the pathway. By focusing on LasI inhibition, our research goal is to develop an antivirulence strategy against MDR strains of P. aeruginosa based on disrupting LasI/LasR quorum sensing pathway using small molecule inhibitors of LasI. This is expected to reduce AHL synthesis, induction of the LasI/LasR pathway, and subsequent virulent behavior. By simulating LuxI-type AHL synthases and docking them with AHL-like inhibitor analogs, we identify a subset of small molecule inhibitors for AHL synthases that may be useful in mitigating virulence in MDR strains of P. aeruginosa.

Potts models and variational autoencoders (VAEs) have recently gained popularity as generative protein sequence models (GPSMs) to explore fitness landscapes and predict the effect of mutations. Despite encouraging results, quantitative characterization and comparison of GPSM-generated probability distributions is still lacking. It is currently unclear whether GPSMs can faithfully reproduce the complex multi-residue mutation patterns observed in natural sequences arising due to epistasis. We develop a set of sequence statistics to assess the \"generative capacity\" of three GPSMs of recent interest: the pairwise Potts Hamiltonian, the VAE, and the site-independent model, using natural and synthetic datasets. We show that the generative capacity of the Potts Hamiltonian model is the largest, in that the higher order mutational statistics generated by the model agree with those observed for natural sequences. In contrast, we show that the VAE's generative capacity lies between the pairwise Potts and site-independent models. Importantly, our work measures GPSM generative capacity in terms of higher-order sequence covariation statistics which we have developed, and provides a new framework for evaluating and interpreting GPSM accuracy that emphasizes the role of epistasis.

Our aim was to investigate whether a nurse-coordinated multidisciplinary, family-based preventive cardiology programme could improve standards of preventive care in routine clinical practice. In a matched, cluster-randomised, controlled trial in eight European countries, six pairs of hospitals and six pairs of general practices were assigned to an intervention programme (INT) or usual care (UC) for patients with coronary heart disease or those at high risk of developing cardiovascular disease. The primary endpoints—measured at 1 year—were family-based lifestyle change; management of blood pressure, lipids, and blood glucose to target concentrations; and prescription of cardioprotective drugs. Analysis was by intention to treat. The trial is registered as ISRCTN 71715857. 1589 and 1499 patients with coronary heart disease in hospitals and 1189 and 1128 at high risk were assigned to INT and UC, respectively. In patients with coronary heart disease who smoked in the month before the event, 136 (58%) in the INT and 154 (47%) in the UC groups did not smoke 1 year afterwards (difference in change 10·4%, 95% CI −0·3 to 21·2, p=0·06). Reduced consumption of saturated fat (196 [55%] vs 168 [40%]; 17·3%, 6·4 to 28·2, p=0·009), and increased consumption of fruit and vegetables (680 [72%] vs 349 [35%]; 37·3%, 18·1 to 56·5, p=0·004), and oily fish (156 [17%] vs 81 [8%]; 8·9%, 0·3 to 17·5, p=0·04) at 1 year were greatest in the INT group. High-risk individuals and partners showed changes only for fruit and vegetables (p=0·005). Blood-pressure target of less than 140/90 mm Hg was attained by both coronary (615 [65%] vs 547 [55%]; 10·4%, 0·6 to 20·2, p=0·04) and high-risk (586 [58%] vs 407 [41%]; 16·9%, 2·0 to 31·8, p=0·03) patients in the INT groups. Achievement of total cholesterol of less than 5 mmol/L did not differ between groups, but in high-risk patients the difference in change from baseline to 1 year was 12·7% (2·4 to 23·0, p=0·02) in favour of INT. In the hospital group, prescriptions for statins were higher in the INT group (810 [86%] vs 794 [80%]; 6·0%, −0·5 to 11·5, p=0·04). In general practices in the intervention groups, angiotensin-converting enzyme inhibitors (297 [29%] INT vs 196 [20%] UC; 8·5%, 1·8 to 15·2, p=0·02) and statins (381 [37%] INT vs 232 [22%] UC; 14·6%, 2·5 to 26·7, p=0·03) were more frequently prescribed. To achieve the potential for cardiovascular prevention, we need local preventive cardiology programmes adapted to individual countries, which are accessible by all hospitals and general practices caring for coronary and high-risk patients. European Society of Cardiology through an unconditional educational grant from AstraZeneca.