Search for: All records

In this paper, we investigate whether symbolic semantic representations, extracted from deep semantic parsers, can help to reason over the states of involved entities in a procedural text. We consider a deep semantic parser (TRIPS) and semantic role labeling as two sources of semantic parsing knowledge. First, we propose PROPOLIS, a symbolic parsing-based procedural reasoning framework. Second, we integrate semantic parsing information into state-of-the-art neural models for procedural reasoning. Our experiments indicate that explicitly incorporating such semantic knowledge improves procedural understanding. This paper presents new metrics for evaluating procedural reasoning tasks that clarify the challenges and identify differences among neural, symbolic, and integrated models. 

The index of refraction (n) of particles is an important parameter in optical models that aims to extract particle size and carbon concentrations from light scattering measurements. An inadequate choice ofncan critically affect the characterization and interpretation of optically-derived parameters, including those from satellite-based models which provide the current view of how biogeochemical processes vary over the global ocean. Yet, little is known about hownvaries over time and space to inform such models. Particularly, in situ estimates ofnfor bulk water samples and at diel-resolving time scales are rare. Here, we demonstrate a method to estimatenusing simultaneously and independently collected particulate beam attenuation coefficients, particle size distribution data, and a Mie theory model. We apply this method to surface waters of the North Pacific Subtropical Gyre (NPSG) at hourly resolution. Clear diel cycles innwere observed, marked by minima around local sunrise and maxima around sunset, qualitatively consistent with several laboratory-based estimates ofnfor specific phytoplankton species. A sensitivity analysis showed that the daily oscillation innamplitude was somewhat insensitive to broad variations in method assumptions, ranging from 11.3 ± 4.3% to 16.9 ± 2.9%. Such estimates are crucial for improvement of algorithms that extract the particle size and production from bulk optical measurements, and could potentially help establish a link betweennvariations and changes in cellular composition of in situ particles.

 

3‐Aminopiperidines are a valuable motif present in small molecule pharmaceuticals and bioactive natural products. Synthesis of these moieties via olefin diamination would be an attractive approach, however significant challenges remain with regards to both regioselectivity and exogenous nucleophile scope. Herein, we report a metal‐free olefin diamination via a “heterocyclic group transfer“ (HGT) reaction of I(III)N‐HVI reagents, giving rise to 3‐aminopiperidines with high selectivity. The HGT strategy leverages heteroarenes as oxidatively masked amine nucleophiles, giving rise to (hetero)arylonium salt products which are isolated via simple trituration and provide a versatile handle for downstream diversification. This represents only the second6‐endoselective I(III)‐mediated diamination reaction and mechanistic studies indicate ring opening of an intermediate aziridinium ion is responsible for the6‐endoselectivity.

 

Ensembl Genomes (https://www.ensemblgenomes.org) provides access to non-vertebrate genomes and analysis complementing vertebrate resources developed by the Ensembl project (https://www.ensembl.org). The two resources collectively present genome annotation through a consistent set of interfaces spanning the tree of life presenting genome sequence, annotation, variation, transcriptomic data and comparative analysis. Here, we present our largest increase in plant, metazoan and fungal genomes since the project's inception creating one of the world's most comprehensive genomic resources and describe our efforts to reduce genome redundancy in our Bacteria portal. We detail our new efforts in gene annotation, our emerging support for pangenome analysis, our efforts to accelerate data dissemination through the Ensembl Rapid Release resource and our new AlphaFold visualization. Finally, we present details of our future plans including updates on our integration with Ensembl, and how we plan to improve our support for the microbial research community. Software and data are made available without restriction via our website, online tools platform and programmatic interfaces (available under an Apache 2.0 license). Data updates are synchronised with Ensembl's release cycle.