Search for: All records

Abstract Accurate prediction of postoperative complications can inform shared decisions regarding prognosis, preoperative risk-reduction, and postoperative resource use. We hypothesized that multi-task deep learning models would outperform conventional machine learning models in predicting postoperative complications, and that integrating high-resolution intraoperative physiological time series would result in more granular and personalized health representations that would improve prognostication compared to preoperative predictions. In a longitudinal cohort study of 56,242 patients undergoing 67,481 inpatient surgical procedures at a university medical center, we compared deep learning models with random forests and XGBoost for predicting nine common postoperative complications using preoperative, intraoperative, and perioperative patient data. Our study indicated several significant results across experimental settings that suggest the utility of deep learning for capturing more precise representations of patient health for augmented surgical decision support. Multi-task learning improved efficiency by reducing computational resources without compromising predictive performance. Integrated gradients interpretability mechanisms identified potentially modifiable risk factors for each complication. Monte Carlo dropout methods provided a quantitative measure of prediction uncertainty that has the potential to enhance clinical trust. Multi-task learning, interpretability mechanisms, and uncertainty metrics demonstrated potential to facilitate effective clinical implementation.

The parasitoid wasp Venturia canescens is an important biological control agent of stored products moth pests and serves as a model to study the function and evolution of domesticated endogenous viruses (DEVs). The DEVs discovered in V. canescens are known as virus-like particles (VcVLPs), which are produced using nudivirus-derived components and incorporate wasp-derived virulence proteins instead of packaged nucleic acids. Previous studies of virus-derived components in the V. canescens genome identified 53 nudivirus-like genes organized in six gene clusters and several viral pseudogenes, but how VcVLP genes are organized among wasp chromosomes following their integration in the ancestral wasp genome is largely unknown. Here, we present a chromosomal scale genome of V. canescens consisting of 11 chromosomes and 56 unplaced small scaffolds. The genome size is 290.8 Mbp with a N50 scaffold size of 24.99 Mbp. A high-quality gene set including 11,831 protein-coding genes were produced using RNA-Seq data as well as publicly available peptide sequences from related Hymenoptera. A manual annotation of genes of viral origin produced 61 intact and 19 pseudogenized nudivirus-derived genes. The genome assembly revealed that two previously identified clusters were joined into a single cluster and a total of 5 gene clusters comprising of 60 intactmore »nudivirus-derived genes were located in three chromosomes. In contrast, pseudogenes are dispersed among 8 chromosomes with only 4 pseudogenes associated with nudivirus gene clusters. The architecture of genes encoding VcVLP components suggests it originates from a recent virus acquisition and there is a link between the processes of dispersal and pseudogenization. This high-quality genome assembly and annotation represents the first chromosome-scale assembly for parasitoid wasps associated with VLPs, and is publicly available in the National Center for Biotechnology Information Genome and RefSeq databases, providing a valuable resource for future studies of DEVs in parasitoid wasps.

« less

Chemical modifications to protein encoding messenger RNAs (mRNAs) influence their localization, translation, and stability within cells. Over 15 different types of mRNA modifications have been observed by sequencing and liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) approaches. While LC-MS/MS is arguably the most essential tool available for studying analogous protein post-translational modifications, the high-throughput discovery and quantitative characterization of mRNA modifications by LC-MS/MS has been hampered by the difficulty of obtaining sufficient quantities of pure mRNA and limited sensitivities for modified nucleosides. We have overcome these challenges by improving the mRNA purification and LC-MS/MS pipelines. The methodologies we developed result in no detectable non-coding RNA modifications signals in our purified mRNA samples, quantify 50 ribonucleosides in a single analysis, and provide the lowest limit of detection reported for ribonucleoside modification LC-MS/MS analyses. These advancements enabled the detection and quantification of 13 S. cerevisiae mRNA ribonucleoside modifications and reveal the presence of four new S. cerevisiae mRNA modifications at low to moderate levels (1-methyguanosine, N 2-methylguanosine, N 2, N 2-dimethylguanosine, and 5-methyluridine). We identified four enzymes that incorporate these modifications into S. cerevisiae mRNAs (Trm10, Trm11, Trm1, and Trm2, respectively), though our results suggest that guanosine and uridine nucleobases aremore »also non-enzymatically methylated at low levels. Regardless of whether they are incorporated in a programmed manner or as the result of RNA damage, we reasoned that the ribosome will encounter the modifications that we detect in cells. To evaluate this possibility, we used a reconstituted translation system to investigate the consequences of modifications on translation elongation. Our findings demonstrate that the introduction of 1-methyguanosine, N 2-methylguanosine and 5-methyluridine into mRNA codons impedes amino acid addition in a position dependent manner. This work expands the repertoire of nucleoside modifications that the ribosome must decode in S. cerevisiae. Additionally, it highlights the challenge of predicting the effect of discrete modified mRNA sites on translation de novo because individual modifications influence translation differently depending on mRNA sequence context.« less

Hematite (α-Fe 2 O 3 ) is a promising transition metal oxide for various energy conversion and storage applications due to its advantages of low cost, high abundance, and good chemical stability. However, its low carrier mobility and electrical conductivity have hindered the wide application of hematite-based devices. Fundamentally, this is mainly caused by the formation of small polarons, which show conduction through thermally activated hopping. Atomic doping is one of the most promising approaches for improving the electrical conductivity in hematite. However, its impact on the carrier mobility and electrical conductivity of hematite at the atomic level remains to be illusive. In this work, through a kinetic Monte-Carlo sampling approach for diffusion coefficients combined with carrier concentrations computed under charge neutrality conditions, we obtained the electrical conductivity of the doped hematite. We considered the contributions from individual Fe–O layers, given that the in-plane carrier transport dominates. We then studied how different dopants impact the carrier mobility in hematite using Sn, Ti, and Nb as prototypical examples. We found that the carrier mobility change is closely correlated with the local distortion of Fe–Fe pairs, i.e. the more stretched the Fe–Fe pairs are compared to the pristine systems, the lower themore »carrier mobility will be. Therefore, elements which limit the distortion of Fe–Fe pair distances from pristine are more desired for higher carrier mobility in hematite. The calculated local structure and pair distribution functions of the doped systems have remarkable agreement with the experimental EXAFS measurements on hematite nanowires, which further validates our first-principles predictions. Our work revealed how dopants impact the carrier mobility and electrical conductivity of hematite and provided practical guidelines to experimentalists on the choice of dopants for the optimal electrical conductivity of hematite and the performance of hematite-based devices.« less