Effective molecular representation learning is of great importance to facilitate molecular property prediction. Recent advances for molecular representation learning have shown great promise in applying graph neural networks to model molecules. Moreover, a few recent studies design self-supervised learning methods for molecular representation to address insufficient labelled molecules; however, these self-supervised frameworks treat the molecules as topological graphs without fully utilizing the molecular geometry information. The molecular geometry, also known as the three-dimensional spatial structure of a molecule, is critical for determining molecular properties. To this end, we propose a novel geometry-enhanced molecular representation learning method (GEM). The proposed GEM has a specially designed geometry-based graph neural network architecture as well as several dedicated geometry-level self-supervised learning strategies to learn the molecular geometry knowledge. We compare GEM with various state-of-the-art baselines on different benchmarks and show that it can considerably outperform them all, demonstrating the superiority of the proposed method.

The sinus node is a collection of highly specialised cells constituting the heart’s pacemaker. The molecular underpinnings of its pacemaking abilities are debated. Using high-resolution mass spectrometry, we here quantify >7,000 proteins from sinus node and neighbouring atrial muscle. Abundances of 575 proteins differ between the two tissues. By performing single-nucleus RNA sequencing of sinus node biopsies, we attribute measured protein abundances to specific cell types. The data reveal significant differences in ion channels responsible for the membrane clock, but not in Ca 2+ clock proteins, suggesting that the membrane clock underpins pacemaking. Consistently, incorporation of ion channel expression differences into a biophysically-detailed atrial action potential model result in pacemaking and a sinus node-like action potential. Combining our quantitative proteomics data with computational modeling, we estimate ion channel copy numbers for sinus node myocytes. Our findings provide detailed insights into the unique molecular make-up of the cardiac pacemaker.

No abstract

Obtaining precise whole-heart segmentation from computed tomography (CT) or other imaging techniques is prerequisite to clinically analyze the cardiac status, which plays an important role in the treatment of cardiovascular diseases. However, the whole-heart segmentation is still a challenging task due to the characteristic of medical images, such as far more background voxels than foreground voxels and the indistinct boundaries of adjacent tissues. In this paper, we first present a new deeply supervised 3D UNET which applies multi-depth fusion to the original network for a better extract context information. Then, we apply focal loss to the field of image segmentation and expand its application to multi-category tasks. Finally, the focal loss is incorporated into the Dice loss function (which can be used to solve category imbalance problem) to form a new loss function, which we call hybrid loss. We evaluate our new pipeline on the MICCAI 2017 whole-heart CT dataset, and it obtains a Dice score of 90.73%, which is better than most of the state-of-the-art methods. INDEX TERMS CT image segmentation, focal loss, deeply-supervised, multi-depth fusion.

Cardiac electrical alternans (CEA), manifested as T-wave alternans in ECG, is a clinical biomarker for predicting cardiac arrhythmias and sudden death. However, the mechanism underlying the spontaneous transition from CEA to arrhythmias remains incompletely elucidated. In this study, multiscale rabbit ventricular models were used to study the transition and a potential role of INa in perpetuating such a transition. It was shown CEA evolved into either concordant or discordant action potential (AP) conduction alternans in a homogeneous one-dimensional tissue model, depending on tissue AP duration and conduction velocity (CV) restitution properties. Discordant alternans was able to cause conduction failure in the model, which was promoted by impaired sodium channel with either a reduced or increased channel current. In a two-dimensional homogeneous tissue model, a combined effect of rate- and curvature-dependent CV broke-up alternating wavefronts at localised points, facilitating a spontaneous transition from CEA to re-entry. Tissue inhomogeneity or anisotropy further promoted break-up of re-entry, leading to multiple wavelets. Similar observations have also been seen in human atrial cellular and tissue models. In conclusion, our results identify a mechanism by which CEA spontaneously evolves into re-entry without a requirement for premature ventricular complexes or pre-existing tissue heterogeneities, and demonstrated the important pro-arrhythmic role of impaired sodium channel activity. These findings are model-independent and have potential human relevance.