Explore the vast range of titles available.

To understand how molecules function in biological systems, new methods are required to obtain atomic resolution structures from biological material under physiological conditions. Intense femtosecond-duration pulses fromX-ray free-electron lasers (XFELs) can outrun most damage processes, vastly increasing the tolerable dose before the specimen is destroyed. This in turn allows structure determination from crystals much smaller and more radiation sensitive than previously considered possible, allowing data collection from room temperature structures and avoiding structural changes due to cooling. Regardless, high-resolution structures obtained from XFEL data mostly use crystals far larger than 1 μm³ in volume, whereas the X-ray beam is often attenuated to protect the detector from damage caused by intense Bragg spots. Here, we describe the 2 Å resolution structure of native nanocrystalline granulovirus occlusion bodies (OBs) that are less than 0.016 μm³ in volume using the full power of the Linac Coherent Light Source (LCLS) and a dose up to 1.3 GGy per crystal. The crystalline shell of granulovirus OBs consists, on average, of about 9,000 unit cells, representing the smallest protein crystals to yield a high-resolution structure by X-ray crystallography to date. The XFEL structure shows little to no evidence of radiation damage and is more complete than a model determined using synchrotron data from recombinantly produced, much larger, cryocooled granulovirus granulin microcrystals. Our measurements suggest that it should be possible, under ideal experimental conditions, to obtain data from protein crystals with only 100 unit cells in volume using currently available XFELs and suggest that single-molecule imaging of individual biomolecules could almost be within reach.

Background Women in Nigeria face some of the highest maternal mortality risks in the world. We explore the benefits and cost-effectiveness of individual and integrated packages of interventions to prevent pregnancy-related deaths. Methods We adapt a previously validated maternal mortality model to Nigeria. Model outcomes included clinical events, population measures, costs, and cost-effectiveness ratios. Separate models were adapted to Southwest and Northeast zones using survey-based data. Strategies consisted of improving coverage of effective interventions, and could include improved logistics. Results Increasing family planning was the most effective individual intervention to reduce pregnancy-related mortality, was cost saving in the Southwest zone and cost-effective elsewhere, and prevented nearly 1 in 5 abortion-related deaths. However, with a singular focus on family planning and safe abortion, mortality reduction would plateau below MDG 5. Strategies that could prevent 4 out of 5 maternal deaths included an integrated and stepwise approach that includes increased skilled deliveries, facility births, access to antenatal/postpartum care, improved recognition of referral need, transport, and availability quality of EmOC in addition to family planning and safe abortion. The economic benefits of these strategies ranged from being cost-saving to having incremental cost-effectiveness ratios less than $500 per YLS, well below Nigeria’s per capita GDP. Conclusions Early intensive efforts to improve family planning and control of fertility choices, accompanied by a stepwise effort to scale-up capacity for integrated maternal health services over several years, will save lives and provide equal or greater value than many public health interventions we consider among the most cost-effective (e.g., childhood immunization).

There is a wide diversity of opinions regarding the management of delayed inflammatory reactions (DIRs) secondary to hyaluronic acid (HA)-based fillers. The plethora of approaches has led the authors to conduct a review regarding management and treatment of DIRs as well as establish therapeutic guidelines for this purpose. A review of the literature was performed through databases such as PubMed using keywords including HA-fillers and complications, delayed HA filler sequelae and therapy, soft tissue and dermal filler reactions and management. Additionally, a survey comprised of questions regarding the management and treatment of DIRs was sent to 18 physicians highly experienced with soft-tissue filler injections in 10 countries. Their answers and recommendations were analyzed and debated amongst these panelists. Sixteen panelists favored antibiotic therapy as first-line treatment for DIRs, specifically dual antibiotic therapy consisting of a fluoroquinolone along with a tetracycline or macrolide for a period of 3-6 weeks. The majority refrained from the use of intralesional (IL) or systemic steroids except in the case of disfiguring or recalcitrant reactions. IL hyaluronidase was recommended by 13 panelists; however, some preferred a watchful waiting approach for a period of 48 hours to 2 weeks prior to IL hyaluronidase, and in cases where antibiotics did not lead to improvement. A consensus was reached and summarized to propose a clear, easy-to-follow, stepwise algorithm for the treatment of DIRs.

Runtime and scalability of large neural networks can be significantly affected by the placement of operations in their dataflow graphs on suitable devices. With increasingly complex neural network architectures and heterogeneous device characteristics, finding a reasonable placement is extremely challenging even for domain experts. Most existing automated device placement approaches are impractical due to the significant amount of compute required and their inability to generalize to new, previously held-out graphs. To address both limitations, we propose an efficient end-to-end method based on a scalable sequential attention mechanism over a graph neural network that is transferable to new graphs. On a diverse set of representative deep learning models, including Inception-v3, AmoebaNet, Transformer-XL, and WaveNet, our method on average achieves 16% improvement over human experts and 9.2% improvement over the prior art with 15 times faster convergence. To further reduce the computation cost, we pre-train the policy network on a set of dataflow graphs and use a superposition network to fine-tune it on each individual graph, achieving state-of-the-art performance on large hold-out graphs with over 50k nodes, such as an 8-layer GNMT.

To the Editor: Wood et al. (May 5 issue) 1 report the results of a randomized study of different doses and levels of dose intensity (dose per unit of time) of adjuvant chemotherapy in women with stage II, node-positive breast cancer. The study is an excellent example of the merits of the conventional study design, but also of its limitations. Wood et al. correctly conclude that the toxicity of treatment is directly related to the dose intensity, the dose rate, or both (their Table 1) 1 . This means that the toxicity is related to the area under the concentration-time curve after . . .