E fficient and cost-effective electrocatalysts play critical roles in energy conversion and storage [1][2][3] . Homogeneous and heterogeneous catalysts represent two parallel frontiers of electrocatalysts, each with their own merits and drawbacks 4,5 . Homogeneous catalysts are attractive for their highly uniform active sites, tunable coordination environment and maximized atom utilization efficiency, but are limited by their relatively poor stability and recyclability. Heterogeneous catalysts are appealing for their high durability, excellent recyclability, and easy immobilization and integration with electrodes, but usually have rather low atom utilization efficiency due to the limited surface sites accessible to reactants. To this end, considerable efforts have been devoted to developing nanoscale heterogeneous catalysts that can increase the exposed surface atoms 3 . However, the inhomogeneity in the distribution of particle sizes and facets poses a serious challenge for controlling active sites and fundamental mechanistic studies 6,7 . In contrast, homogeneous catalysts typically exhibit the well-defined atomic structure with tunable coordination environment that is essential for deciphering the catalytic reaction pathway and rational design of targeted catalysts with tailored catalytic properties 8 . Single-atom catalysts (SACs) with monodispersed single atoms supported on solid substrates are recently emerging as an exciting class of catalysts that combine the merits of both homogeneous and heterogeneous catalysts [9][10][11][12][13][14] . However, most SACs studied to date employ metal oxides (for example, TiO 2 , CeO 2 and FeO x ) as supporting substrates to prevent atom aggregation [15][16][17][18] , which cannot be readily applied in electrocatalytic applications due to their low electrical conductivity and/or poor stability in harsh liquid-phase electrolytes (for example, strong acid or base). Atomic transitionmetal-nitrogen moieties supported in carbon (M-N-Cs) represent a unique class of SACs with high electrical conductivity and superior (electro)chemical stability for electrocatalytic applications 19 . In particular, Fe-based M-N-Cs have been extensively studied as electrocatalysts towards the oxygen reduction reaction (ORR) with demonstrated activity and stability approaching those of commercial Pt/C catalysts 20,21 . In addition, as suggested by numerous theoretical studies, M-N-Cs are promising candidates for catalysing a wide range of electrochemical processes, such as hydrogen reduction/oxidation 22 , CO 2 /CO reduction 23 and N 2 reduction 24 . A significant advantage of SACs is that the well-defined single atomic site could allow precise understanding of the catalytic reaction pathway, and rational design of targeted catalysts with tailored activity (in a manner similar to homogeneous catalyst design). However, this perceived advantage has been investigated theoretically

miRDeepFinder is a software package developed to identify and functionally analyze plant microRNAs (miRNAs) and their targets from small RNA datasets obtained from deep sequencing. The functions available in miRDeepFinder include pre-processing of raw data, identifying conserved miRNAs, mining and classifying novel miRNAs, miRNA expression profiling, predicting miRNA targets, and gene pathway and gene network analysis involving miRNAs. The fundamental design of miRDeepFinder is based on miRNA biogenesis, miRNA-mediated gene regulation and target recognition, such as perfect or near perfect hairpin structures, different read abundances of miRNA and miRNA*, and targeting patterns of plant miRNAs. To test the accuracy and robustness of miRDeepFinder, we analyzed a small RNA deep sequencing dataset of Arabidopsis thaliana published in the GEO database of NCBI. Our test retrieved 128 of 131 (97.7%) known miRNAs that have a more than 3 read count in Arabidopsis. Because many known miRNAs are not associated with miRNA*s in small RNA datasets, miRDeepFinder was also designed to recover miRNA candidates without the presence of miRNA*. To mine as many miRNAs as possible, miRDeepFinder allows users to compare mature miRNAs and their miRNA*s with other small RNA datasets from the same species. Cleaveland software package was also incorporated into miRDeepFinder for miRNA target identification using degradome sequencing analysis. Using this new computational tool, we identified 13 novel miRNA candidates with miRNA*s from Arabidopsis and validated 12 of them experimentally. Interestingly, of the 12 verified novel miRNAs, a miRNA named AC1 spans the exons of two genes (UTG71C4 and UGT71C3). Both the mature AC1 miRNA and its miRNA* were also found in four other small RNA datasets. We also developed a tool, "miRNA primer designer" to design primers for any type of miRNAs. miRDeepFinder provides a powerful tool for analyzing small RNA datasets from all species, with or without the availability of genome information. miRDeepFinder and miRNA primer designer are freely available at http://www.leonxie.com/DeepFinder.php and at http://www.leonxie.com/miRNAprimerDesigner.php , respectively. A program (called RefFinder: http://www.leonxie.com/referencegene.php ) was also developed for assessing the reliable reference genes for gene expression analysis, including miRNAs.

The cathodic oxygen reduction reaction (ORR) is essential in the electrochemical energy conversion of fuel cells. Here, through the NH atmosphere annealing of a graphene oxide (GO) precursor containing trace amounts of Ru, we have synthesized atomically dispersed Ru on nitrogen-doped graphene that performs as an electrocatalyst for the ORR in acidic medium. The Ru/nitrogen-doped GO catalyst exhibits excellent four-electron ORR activity, offering onset and half-wave potentials of 0.89 and 0.75 V, respectively, vs a reversible hydrogen electrode (RHE) in 0.1 M HClO, together with better durability and tolerance toward methanol and carbon monoxide poisoning than seen in commercial Pt/C catalysts. X-ray adsorption fine structure analysis and aberration-corrected high-angle annular dark-field scanning transmission electron microscopy are performed and indicate that the chemical structure of Ru is predominantly composed of isolated Ru atoms coordinated with nitrogen atoms on the graphene substrate. Furthermore, a density function theory study of the ORR mechanism suggests that a Ru-oxo-N structure appears to be responsible for the ORR catalytic activity in the acidic medium. These findings provide a route for the design of efficient ORR single-atom catalysts.