In this work, we analyzed a data set formed by 566 donor/acceptor pairs, which are part of organic solar cells recently reported. We explored the effect of different descriptors in machine learning (ML) models to predict the power conversion efficiency (PCE) of these cells. The investigated descriptors are classified into two main categories: structural (topology properties) and physical descriptors (energy levels, molecular size, light absorption, and mixing properties). In line with previous observations, ML predictions are more accurate when using both structural and physical descriptors, as opposed to only using one of them. We observed that ML predictions are also improved by using larger and more varied data sets. Importantly, the structural descriptors are the ones contributing the most to the ML models. Some physical properties are highly correlated with PCE, although they do not improve notably the ML prediction accuracy as they carry information already encoded in the structural descriptors. Given that various descriptors have significantly different computational costs, the analysis presented here can be used as a guide to construct ML models that maximize predictive power and minimize computational costs for screening large sets of candidates.

Voltage-gated sodium channel Nav1.5 has been linked to the cardiac cell excitability and a variety of arrhythmic syndromes including long QT, Brugada, and conduction abnormalities. Nav1.5 exhibits a slow inactivation, corresponding to a duration-dependent bi-exponential recovery, which is often associated with various arrhythmia syndromes. However, the gating mechanism of Nav1.5 and the physiological role of slow inactivation in cardiac cells remain elusive. Here a 12-state two-step inactivation Markov model was successfully developed to depict the gating kinetics of Nav1.5. This model can simulate the Nav1.5 channel in not only steady state processes, but also various transient processes. Compared with the simpler 8-state model, this 12-state model is well-behaved in simulating and explaining the processes of slow inactivation and slow recovery. This model provides a good framework for further studying the gating mechanism and physiological role of sodium channel in excitable cells.

Cytosolic protein delivery is a prerequisite for protein‐based biotechnologies and therapeutics on intracellular targets. Polymers that can complex with proteins to form nano‐assemblies represent one of the most important categories of materials, because of the ease of nano‐fabrication, high protein loading efficiency, no need for purification, and maintenance of protein bioactivity. Stable protein encapsulation and efficient intracellular liberation are two critical yet opposite processes toward cytosolic delivery, and polymers that can resolve these two conflicting challenges are still lacking. Herein, hyperbranched poly(β‐amino ester) (HPAE) with backbone‐embedded phenylboronic acid (PBA) is developed to synchronize these two processes, wherein PBA enhanced protein encapsulation via nitrogen–boronate (N–B) coordination while triggered polymer degradation and protein release upon oxidation by H2O2 in cancer cells. Upon optimization of the branching degree, charge density, and PBA distribution, the best‐performing A2‐B3‐C2‐S2‐P2 is identified, which mediates robust delivery of various native proteins/peptides with distinct molecular weights (1.6–430 kDa) and isoelectric points (4.1–10.3) into cancer cells, including enzymes, toxins, antibodies, and CRISPR‐Cas9 ribonucleoproteins (RNPs). Moreover, A2‐B3‐C2‐S2‐P2 mediates effective cytosolic delivery of saporin both in vitro and in vivo to provoke remarkable anti‐tumor efficacy. Such a potent and universal platform holds transformative potentials for protein pharmaceuticals.

Biomimetic nanodecoys reconstruct the osteoclast/osteoblast balance to treat postmenopausal osteoporosis.