Explore the vast range of titles available.

As interest in unsupervised deep learning models for the analysis of gene expression data has grown, an increasing number of methods have been developed to make these deep learning models more interpretable. These methods can be separated into two groups: (1) post hoc analyses of black box models through feature attribution methods and (2) approaches to build inherently interpretable models through biologically-constrained architectures. In this work, we argue that these approaches are not mutually exclusive, but can in fact be usefully combined. We propose a novel unsupervised pathway attribution method, which better identifies major sources of transcriptomic variation than prior methods when combined with biologically-constrained neural network models. We demonstrate how principled feature attributions aid in the analysis of a variety of single cell datasets. Finally, we apply our approach to a large dataset of post-mortem brain samples from patients with Alzheimer's disease, and show that it identifies Mitochondrial Respiratory Complex I as an important factor in this disease. Competing Interest Statement The authors have declared no competing interest.

Identifying gene expression markers for Alzheimer's disease (AD) neuropathology through meta-analysis is a complex undertaking because available data are often from different studies and/or brain regions involving study-specific confounders and/or region-specific biological processes. Here, we developed a probabilistic model-based framework, DECODER, leveraging these discrepancies to identify robust biomarkers for complex phenotypes. Our experiments present: (1) DECODER's potential as a general meta-analysis framework widely applicable to various diseases (e.g., AD and cancer) and phenotypes (e.g., Amyloid-beta (Abeta) pathology, tau pathology, and survival), (2) our results from a meta-analysis using 1,746 human brain tissue samples from nine brain regions in three studies -- the largest expression meta-analysis for AD, to our knowledge --, and (3) in vivo validation of identified modifiers of Abeta toxicity in a transgenic Caenorhabditis elegans model expressing AD-associated Abeta, which pinpoints mitochondrial Complex I as a critical mediator of proteostasis and a promising pharmacological avenue toward treating AD.

In some organisms, miRNA-mediated regulation leads to RNA-dependent RNA polymerase based amplification, involving generation of secondary siRNAs that target the same transcript, but why this happens for some miRNAs and not others has been unclear. It is now shown that Arabidopsis miRNAs that trigger this amplified response tend to be 22 nt rather than 21 nt in length, and that generation of 22 nt miRNAs this length seems to depend on the foldback structure of the precursor. RNA interference pathways can involve amplification of secondary siRNAs by RNA-dependent RNA polymerases. In plants, RDR6-dependent secondary siRNAs arise from transcripts targeted by some microRNAs (miRNAs). Here, Arabidopsis thaliana secondary siRNAs from mRNA as well as trans -acting siRNAs are shown to be triggered through initial targeting by a 22-nucleotide (nt) miRNA that associates with AGO1. In contrast to canonical 21-nt miRNAs, 22-nt miRNAs primarily arise from foldback precursors containing asymmetric bulges. Using artificial miRNA constructs, conversion of asymmetric foldbacks to symmetric foldbacks resulted in the production of 21-nt forms of miR173, miR472 and miR828. Both 21- and 22-nt forms associated with AGO1 and guided accurate slicer activity, but only 22-nt forms were competent to trigger RDR6-dependent siRNA production from target RNA. These data suggest that AGO1 functions differentially with 21- and 22-nt miRNAs to engage the RDR6-associated amplification apparatus.

Invasive mammals are the greatest threat to island biodiversity and invasive rodents are likely responsible for the greatest number of extinctions and ecosystem changes. Techniques for eradicating rodents from islands were developed over 2 decades ago. Since that time there has been a significant development and application of this conservation tool. We reviewed the literature on invasive rodent eradications to assess its current state and identify actions to make it more effective. Worldwide, 332 successful rodent eradications have been undertaken; we identified 35 failed eradications and 20 campaigns of unknown result. Invasive rodents have been eradicated from 284 islands (47,628 ha). With the exception of two small islands, rodenticides were used in all eradication campaigns. Brodifacoum was used in 71% of campaigns and 91% of the total area treated. The most frequent rodenticide distribution methods (from most to least) are bait stations, hand broadcasting, and aerial broadcasting. Nevertheless, campaigns using aerial broadcast made up 76% of the total area treated. Mortality of native vertebrates due to nontarget poisoning has been documented, but affected species quickly recover to pre-eradication population levels or higher. A variety of methods have been developed to mitigate nontarget impacts, and applied research can further aid in minimizing impacts. Land managers should routinely remove invasive rodents from islands <100 ha that lack vertebrates susceptible to nontarget poisoning. For larger islands and those that require nontarget mitigation, expert consultation and greater planning effort are needed. With the exception of house mice (Mus musculus), island size may no longer be the limiting factor for rodent eradications; rather, social acceptance and funding may be the main challenges. To be successful, large-scale rodent campaigns should be integrated with programs to improve the livelihoods of residents, island biosecurity, and reinvasion response programs.

Transacting siRNA (tasiRNA) biogenesis in Arabidopsis is initiated by microRNA (miRNA) -guided cleavage of primary transcripts. In the case of TAS3 tasiRNA formation, ARGONAUTE7 (AGO7)-miR390 complexes interact with primary transcripts at two sites, resulting in recruitment of RNA-DEPENDENT RNA POLYMERASE6 for dsRNA biosynthesis. An extensive screen for Arabidopsis mutants with specific defects in TAS3 tasiRNA biogenesis or function was done. This yielded numerous ago7 mutants, one dcl4 mutant, and two mutants that accumulated low levels of miR390. A direct genome sequencing-based approach to both map and rapidly identify one of the latter mutant alleles was developed. This revealed a G-to-A point mutation (mir390a-1) that was calculated to stabilize a relatively nonpaired region near the base of the MIR390a foldback, resulting in misprocessing of the miR390/miR390* duplex and subsequent reduced TAS3 tasiRNA levels. Directed substitutions, as well as analysis of variation at paralogous miR390-generating loci (MIR390a and MIR390b), indicated that base pair properties and nucleotide identity within a region 4-6 bases below the miR390/miR390* duplex region contributed to the efficiency and accuracy of precursor processing.

The emergence of highly transmissible SARS-CoV-2 variants has created a need to reassess the risk posed by increasing social contacts as countries resume pre-pandemic activities, particularly in the context of resuming large-scale events over multiple days. To examine how social contacts formed in different activity settings influences interventions required to control Delta variant outbreaks, we collected high-resolution data on contacts among passengers and crew on cruise ships and combined the data with network transmission models. We found passengers had a median of 20 (IQR 10–36) unique close contacts per day, and over 60% of their contact episodes were made in dining or sports areas where mask wearing is typically limited. In simulated outbreaks, we found that vaccination coverage and rapid antigen tests had a larger effect than mask mandates alone, indicating the importance of combined interventions against Delta to reduce event risk in the vaccine era. Here, the authors simulate COVID-19 outbreaks on an empirical contact network derived from digital contact data collected on cruise ships. They model impacts of different control measures and find that combinations of measures, particularly vaccination and rapid antigen testing, are important for mitigating outbreaks.

Case isolation and contact tracing can contribute to the control of COVID-19 outbreaks 1 , 2 . However, it remains unclear how real-world social networks could influence the effectiveness and efficiency of such approaches. To address this issue, we simulated control strategies for SARS-CoV-2 transmission in a real-world social network generated from high-resolution GPS data that were gathered in the course of a citizen-science experiment 3 , 4 . We found that tracing the contacts of contacts reduced the size of simulated outbreaks more than tracing of only contacts, but this strategy also resulted in almost half of the local population being quarantined at a single point in time. Testing and releasing non-infectious individuals from quarantine led to increases in outbreak size, suggesting that contact tracing and quarantine might be most effective as a ‘local lockdown’ strategy when contact rates are high. Finally, we estimated that combining physical distancing with contact tracing could enable epidemic control while reducing the number of quarantined individuals. Our findings suggest that targeted tracing and quarantine strategies would be most efficient when combined with other control measures such as physical distancing. Combining fine-scale social contact data with epidemic modeling reveals interactions among contact tracing, quarantine, testing and physical distancing for controlling COVID-19.

Our climate model, driven mainly by increasing human-made greenhouse gases and aerosols, among other forcings, calculates that Earth is now absorbing 0.85 ± 0.15 watts per square meter more energy from the Sun than it is emitting to space. This imbalance is confirmed by precise measurements of increasing ocean heat content over the past 10 years. Implications include (i) the expectation of additional global warming of about 0.6°C without further change of atmospheric composition; (ii) the confirmation of the climate system's lag in responding to forcings, implying the need for anticipatory actions to avoid any specified level of climate change; and (iii) the likelihood of acceleration of ice sheet disintegration and sea level rise.

Bioimaging data have significant potential for reuse, but unlocking this potential requires systematic archiving of data and metadata in public databases. We propose draft metadata guidelines to begin addressing the needs of diverse communities within light and electron microscopy. We hope this publication and the proposed Recommended Metadata for Biological Images (REMBI) will stimulate discussions about their implementation and future extension.

In this study, physical and chemical properties of ultrafine aerosol particles are investigated at an urban site in Bakersfield, California, during the CalNex 2010 (California Research at the Nexus of Air Quality and Climate Change) campaign in May and June. Ultrafine particle measurements include particle number size distributions by a scanning Differential Mobility Analyzer (DMA) and size resolved aerosol chemical composition determined with a High‐Resolution Time‐of‐Flight Aerosol Mass Spectrometer (HR‐ToF‐AMS). Growth events of ultrafine particles were observed on most days and had a very regular pattern. A nucleation mode centered at ∼20 nm appeared in the morning and grew to 40–100 nm throughout the day. Microphysical modeling and size‐resolved HR‐ToF‐AMS concentrations showed that organic components provided most of the particle growth in the ultrafine mode, and sulfate provided on most days only a minor contribution to the mass of this mode. The ultrafine particle mass was largely dominated by organics (77%), and was at maximum during the afternoon. Elemental carbon (EC) and the AMS tracer C4H9+ for hydrocarbon‐like organic aerosol (HOA) peaked in the early morning during rush hour, indicative of primary emissions. The fact that the particle number concentration peaked in the afternoon, when EC was at minimum, indicates that the midday increase in number concentration was likely due to new particle formation. The potential importance of solar radiation, the condensation sink of vapor on existing particles, concentrations of OH, O3, SO2, NH3, and VOCs for both condensational growth and new particle formation is evaluated based on the covariation of these parameters with ultrafine mass. The results suggest that the ultrafine particles are from secondary sources that are co‐emitted or co‐produced with glyoxal and formaldehyde. Key Points Ultrafine particle growth events were dominated by organic components Particle number concentration was controlled by new particle formation Ultrafine particles mass was correlated with formaldehyde and glyoxal