Ribonanza: deep learning of RNA structure through dual crowdsourcing.

Prediction of RNA structure from sequence remains an unsolved problem, and progress has been slowed by a paucity of experimental data. Here, we present Ribonanza, a dataset of chemical mapping measurements on two million diverse RNA sequences collected through Eterna and other crowdsourced initiatives. Ribonanza measurements enabled solicitation, training, and prospective evaluation of diverse deep neural networks through a Kaggle challenge, followed by distillation into a single, self-contained model called RibonanzaNet. When fine tuned on auxiliary datasets, RibonanzaNet achieves state-of-the-art performance in modeling experimental sequence dropout, RNA hydrolytic degradation, and RNA secondary structure, with implications for modeling RNA tertiary structure.

Nucleic Transformer: Classifying DNA Sequences with Self-Attention and Convolutions.

Much work has been done to apply machine learning and deep learning to genomics tasks, but these applications usually require extensive domain knowledge, and the resulting models provide very limited interpretability. Here, we present the Nucleic Transformer, a conceptually simple but effective and interpretable model architecture that excels in the classification of DNA sequences. The Nucleic Transformer employs self-attention and convolutions on nucleic acid sequences, leveraging two prominent deep learning strategies commonly used in computer vision and natural language analysis. We demonstrate that the Nucleic Transformer can be trained without much domain knowledge to achieve high performance in Escherichia coli promoter classification, viral genome identification, enhancer classification, and chromatin profile predictions.

Consistency of limb preference across unimanual feeding, bipedal locomotion, and social grooming in golden snub-nosed monkeys (Rhinopithecus roxellana).

The golden snub-nosed monkey (Rhinopithecus roxellana) is a typical arboreal group-living Old World primate. While limb preference has been extensively studied in this species, limb preference consistency has not yet been explored. Here, based on 26 R. roxellana adults, we investigated whether individuals exhibit consistent motor biases in manual- (e.g., unimanual feeding and social grooming) and foot-related (e.g., bipedal locomotion) tasks and whether limb preference consistency is influenced by increased social interactions during social grooming. Results showed no consistency in the direction or strength of limb preference among tasks, except for lateral strength in handedness for unimanual feeding and footedness in the initiation of locomotion. Population-level foot preference was only found among right-handers. Marked lateral bias was found in unimanual feeding, indicating that it may be a sensitive behavioural measure for assessing manual preference, especially for provisioned populations. This study not only improves our understanding of the relationship between hand and foot preference in R. roxellana but also reveals potential differential hemispheric regulation of limb preference and the influence of increased social interaction on handedness consistency.

Does diet or macronutrients intake drive the structure and function of gut microbiota?

Shift of ingestive behavior is an important strategy for animals to adapt to change of the environment. We knew that shifts in animal dietary habits lead to changes in the structure of the gut microbiota, but we are not sure about if changes in the composition and function of the gut microbiota respond to changes in the nutrient intake or food items. To investigate how animal feeding strategies affect nutrient intakes and thus alter the composition and digestion function of gut microbiota, we selected a group of wild primate group for the study. We quantified their diet and macronutrients intake in four seasons of a year, and instant fecal samples were analyzed by high-throughput sequencing of 16S rRNA and metagenomics. These results demonstrated that the main reason that causes seasonal shifts of gut microbiota is the macronutrient variation induced by seasonal dietary differences. Gut microbes can help to compensate for insufficient macronutrients intake of the host through microbial metabolic functions. This study contributes to a deeper understanding of the causes of seasonal variation in host-microbial variation in wild primates.

RNAdegformer: accurate prediction of mRNA degradation at nucleotide resolution with deep learning.

Messenger RNA-based therapeutics have shown tremendous potential, as demonstrated by the rapid development of messenger RNA based vaccines for COVID-19. Nevertheless, distribution of mRNA vaccines worldwide has been hampered by mRNA's inherent thermal instability due to in-line hydrolysis, a chemical degradation reaction. Therefore, predicting and understanding RNA degradation is a crucial and urgent task. Here we present RNAdegformer, an effective and interpretable model architecture that excels in predicting RNA degradation. RNAdegformer processes RNA sequences with self-attention and convolutions, two deep learning techniques that have proved dominant in the fields of computer vision and natural language processing, while utilizing biophysical features of RNA. We demonstrate that RNAdegformer outperforms previous best methods at predicting degradation properties at nucleotide resolution for COVID-19 mRNA vaccines. RNAdegformer predictions also exhibit improved correlation with RNA in vitro half-life compared with previous best methods. Additionally, we showcase how direct visualization of self-attention maps assists informed decision-making. Further, our model reveals important features in determining mRNA degradation rates via leave-one-feature-out analysis.

Pregnancy diagnosis and sex identification with urinary glycopatterns of two mammal species.

Glycome in urine could be promising biomarkers for detecting pregnancy diagnosis and sex noninvasively for animals, especially for rare species. We explore the applicability of grouping golden snub-nosed monkeys by sex or diagnosing pregnancy based on their urinary glycopatterns, which are determined via lectin microarray combining mass spectrometry analysis. Sprague-Dawley rats are used to verify whether this approach and whether the glycomic biomarkers can be generalized to other mammalian species. The results show that, for both species, lectin microarray combining mass spectrometry can distinguish individuals' pregnancy status and sex; significant differences are found in the types, amounts, and terminal modification of glycans between pregnant and non-pregnant females and between females and males. This indicates the approach could be generalized to other mammalian species to group sex and detect pregnancy, yet the glycopatterns appear to be species-specific and markers developed from one species may not be directly applicable to another.

vcfpop: Performing population genetics analyses for autopolyploids and aneuploids based on next-generation sequencing data sets.

Polyploids are cells or organisms with a genome consisting of more than two sets of homologous chromosomes. Polyploid plants have important traits that facilitate speciation and are thus often model systems for evolutionary, molecular ecology and agricultural studies. However, due to their unusual mode of inheritance and double-reduction, diploid models of population genetic analysis cannot properly be applied to autopolyploids. To overcome this problem, we developed a software package entitled vcfpop to perform a variety of population genetic analyses for autopolyploids, such as parentage analysis, analysis of molecular variance, principal coordinates analysis, hierarchical clustering analysis and Bayesian clustering. We used three data sets to evaluate the capability of vcfpop to analyse large data sets on a desktop computer. This software is freely available at http://github.com/huangkang1987/vcfpop.

Deep learning models for predicting RNA degradation via dual crowdsourcing.

Medicines based on messenger RNA (mRNA) hold immense potential, as evidenced by their rapid deployment as COVID-19 vaccines. However, worldwide distribution of mRNA molecules has been limited by their thermostability, which is fundamentally limited by the intrinsic instability of RNA molecules to a chemical degradation reaction called in-line hydrolysis. Predicting the degradation of an RNA molecule is a key task in designing more stable RNA-based therapeutics. Here, we describe a crowdsourced machine learning competition ('Stanford OpenVaccine') on Kaggle, involving single-nucleotide resolution measurements on 6,043 diverse 102-130-nucleotide RNA constructs that were themselves solicited through crowdsourcing on the RNA design platform Eterna. The entire experiment was completed in less than 6 months, and 41% of nucleotide-level predictions from the winning model were within experimental error of the ground truth measurement. Furthermore, these models generalized to blindly predicting orthogonal degradation data on much longer mRNA molecules (504-1,588 nucleotides) with improved accuracy compared with previously published models. These results indicate that such models can represent in-line hydrolysis with excellent accuracy, supporting their use for designing stabilized messenger RNAs. The integration of two crowdsourcing platforms, one for dataset creation and another for machine learning, may be fruitful for other urgent problems that demand scientific discovery on rapid timescales.

Intraoperative cell salvage as an effective intervention for postpartum hemorrhage-Evidence from a prospective randomized controlled trial.

Background: Postpartum hemorrhage (PPH) is one of the leading causes of maternal mortality. Promptly recovering blood loss is critical for PPH. Intraoperative cell salvage (ICS) is a method to collect and process red blood cells (RBCs) from the blood lost during surgery and transfuse them to the patient's circulation during or immediately after surgery. Its effectiveness in reducing the demand for allogeneic blood transfusion has been proven, but its effectiveness and safety as a sole treatment for PPH during Cesarean sections are unclear. This is particularly important for patients who cannot or do not want to accept allogeneic blood transfusion. Materials and methods: In this prospective randomized controlled study, patients with high risks of PPH were randomized into the ICS group or the control group, receiving ICS or allogeneic RBC transfusion if their hemoglobin level was less than 80 g/L during operation. Data collected include clinical examination, blood cell count, hemoglobin level, coagulation function, and plasma levels of fetal hemoglobin, tissue factor, and alpha-fetoprotein before and after fetal delivery and 0, 2, and 12 h after treatment. Adverse events were recorded. Results: A total of 130 patients were enrolled, aged 33 ± 1 years with a mean gestation period of 37 ± 1 week. The most common cause of Cesarean section was placenta previa, followed by twin pregnancy, scarred uterus, preeclampsia, placental abruption, fetal distress, and placenta accreta spectrum. Bleeding amount was similar between the two groups. The ICS group, compared to controls, had more efficient increases in levels of hemoglobin, RBC, and hematocrit (all p < 0.05). Coagulation function was maintained in the ICS group but reduced in controls 24 h after transfusion, indicated by significantly reduced fibrinogen level and prolonged prothrombin time (PT), thrombin time (TT), and activated partial thromboplastin time (aPTT) (all p < 0.05). There was a transient but significant decrease in plasma tissue factor and alpha-fetoprotein levels and an increase in plasma fetal hemoglobin level with ICS treatment in the postpartum period. No adverse event occurred with ICS intervention. Conclusion: ICS is an effective and safe intervention for patients with a high risk of PPH during elective or emergency Cesarean section. It can effectively clear tissue factors and alpha-fetoprotein but not fetal hemoglobin.

Deep learning models for predicting RNA degradation via dual crowdsourcing.

Messenger RNA-based medicines hold immense potential, as evidenced by their rapid deployment as COVID-19 vaccines. However, worldwide distribution of mRNA molecules has been limited by their thermostability, which is fundamentally limited by the intrinsic instability of RNA molecules to a chemical degradation reaction called in-line hydrolysis. Predicting the degradation of an RNA molecule is a key task in designing more stable RNA-based therapeutics. Here, we describe a crowdsourced machine learning competition ("Stanford OpenVaccine") on Kaggle, involving single-nucleotide resolution measurements on 6043 102-130-nucleotide diverse RNA constructs that were themselves solicited through crowdsourcing on the RNA design platform Eterna. The entire experiment was completed in less than 6 months, and 41% of nucleotide-level predictions from the winning model were within experimental error of the ground truth measurement. Furthermore, these models generalized to blindly predicting orthogonal degradation data on much longer mRNA molecules (504-1588 nucleotides) with improved accuracy compared to previously published models. Top teams integrated natural language processing architectures and data augmentation techniques with predictions from previous dynamic programming models for RNA secondary structure. These results indicate that such models are capable of representing in-line hydrolysis with excellent accuracy, supporting their use for designing stabilized messenger RNAs. The integration of two crowdsourcing platforms, one for data set creation and another for machine learning, may be fruitful for other urgent problems that demand scientific discovery on rapid timescales.

Insights into the interactions of bisphenol and phthalate compounds with unamended and carnitine-amended montmorillonite clays.

Montmorillonite clays could be promising sorbents to mitigate toxic compound exposures. Bisphenols A (BPA) and S (BPS) as well as phthalates, dibutyl phthalate (DBP) and di-2-ethylhexyl phthalate (DEHP), are ubiquitous environmental contaminants linked to adverse health effects. Here, we combined computational and experimental methods to investigate the ability of montmorillonite clays to sorb these compounds. Molecular dynamics simulations predicted that parent, unamended, clay has higher binding propensity for BPA and BPS than for DBP and DEHP; carnitine-amended clay improved BPA and BPS binding, through carnitine simultaneously anchoring to the clay through its quaternary ammonium cation and forming hydrogen bonds with BPA and BPS. Experimental isothermal analysis confirmed that carnitine-amended clay has enhanced BPA binding capacity, affinity and enthalpy. Our studies demonstrate how computational and experimental methods, combined, can characterize clay binding and sorption of toxic compounds, paving the way for future investigation of clays to reduce BPA and BPS exposure.

Male social rank and food competition in a primate multi-level society.

OBJECTIVES: Social animals often have dominance hierarchies, with high rank conferring preferential access to resources. In primates, competition among males is often assumed to occur predominantly over reproductive opportunities. However, competition for food may occur during food shortages, such as in temperate species during winter. Higher-ranked males may thus gain preferential access to high-profitability food, which would enable them to spend longer engaged in activities other than feeding. MATERIALS AND METHODS: We performed a field experiment with a breeding band of golden snub-nosed monkeys, a species that lives in a multi-level society in high-altitude forests in central China. We provisioned monkey's high-profitability food during winter when natural foods are limited, and then recorded the times individual adult males spent engaged in different behaviors. RESULTS: Higher-ranking males spent less time feeding overall and fed on provisioned foods at a higher rate than lower-ranking males. Higher-ranking males therefore had more time to spend on alternative behaviors. We found no significant difference according to rank in times spent moving or resting. However, high-ranking males spend significantly longer on affiliative behaviors with other members of their social sub-units, especially grooming and being groomed, behaviors known to promote social cohesion in primates. DISCUSSION: We show that preferential access to high-profitability foods likely relaxes time-budget constraints to higher-ranking males. High-ranking males thus spend more time on non-feeding activities, especially grooming, which may enhance social cohesion within their social sub-unit. We discuss the potential direct and indirect benefits to high-ranking males associated with preferential access to high-value food during winter.

Observation of mother-perpetrated infanticide in golden takins ( Budorcas taxicolor bedfordi).

Infanticide by unrelated individuals is widely reported in the animal kingdom; however, little is known about cases perpetrated by a parent, particularly the mother. This article reports on three cases of mother-initiated infanticide in Qinling golden takins ( Budorcas taxicolor bedfordi) recorded from video and camera images. Based on previous reports in other animals, we propose that the infanticide events observed in golden takins were related to the parental manipulation mechanism - i.e., killing an unhealthy infant to allow the mother to invest more care in potentially healthy offspring, and gain more fruitful reproductive opportunities. This appears to be an evolutionary-based selection strategy, whereby a species can prosper and succeed under the challenges of natural selection. However, further studies on both captive and wild populations are required to answer the various questions raised from our observations.

Different MicroRNA Families Involved in Regulating High Temperature Stress Response during Cotton (Gossypium hirsutum L.) Anther Development.

MicroRNAs (miRNAs) are small molecule RNAs widely involved in responses to plant abiotic stresses. We performed small RNA sequencing of cotton anthers at four developmental stages under normal and high temperature (NT and HT, respectively) conditions to investigate the stress response characteristics of miRNA to HT. A total of 77 miRNAs, including 33 known miRNAs and 44 novel miRNAs, were identified, and 41 and 28 miRNAs were differentially expressed under NT and HT stress conditions, respectively. The sporogenous cell proliferation (SCP), meiotic phase (MP), microspore release period (MRP), and pollen maturity (PM) stages had 10 (including 12 miRNAs), four (including six miRNAs), four (including five miRNAs), and seven (including 11 miRNAs) HT stress-responsive miRNA families, respectively, which were identified after removing the changes in genotype-specific miRNAs under NT condition. Seven miRNA families (miR2949, miR167, and miR160 at the SCP stage; miR156 and miR172 at the MP stage; miR156 at the MRP stage; and miR393 and miR3476 at the PM stage), which had expression abundance of more than 10% of the total expression abundance, served as the main regulators responding to HT stress with positive or negative regulation patterns. These miRNAs orchestrated the expression of the corresponding target genes and led to different responses in the HT-tolerant and the HT-sensitive lines. The results revealed that the HT stress response of miRNAs in cotton anthers were stage-specific and differed with the development of anthers. Our study may enhance the understanding of the response of miRNAs to HT stress in cotton anthers and may clarify the mechanism of plant tolerance to HT stress.

Montmorillonites Can Tightly Bind Glyphosate and Paraquat Reducing Toxin Exposures and Toxicity.

Among the numerous contaminants of soil, glyphosate and paraquat are two of the most widely used herbicides that are commonly detected in the environment. Soil and sediment contaminated with glyphosate, paraquat, and other environmental toxins can be mobilized and redistributed to lawns, vegetable gardens, parks, and water supplies in vulnerable communities at the site of disasters such as hurricanes and flooding. Glyphosate and paraquat bind strongly to soils containing clays, making their bioavailability (bioaccessibility) from these types of soil very low. Because of their affinity for clay-based soils, it is possible that montmorillonite clays could be administered as a therapeutic agent in the diet of animals and humans to decrease short-term exposure and toxicity. In this study, we investigated the sorption mechanisms of glyphosate and paraquat onto active surfaces of calcium montmorillonite (CM) and sodium montmorillonite (SM) clays and derived binding parameters, including capacity, affinity, and enthalpy. Additionally, we used these parameters to predict the reduction in bioavailability under different pH and temperature conditions and to estimate the theoretical dose of clay that could protect against severe paraquat toxicity and lethality. Computational modeling and simulation studies depicted toxin sorption mechanisms at different pH values. Additionally, a toxin-sensitive living organism (Hydra vulgaris) was used to confirm the safety of the clay and its ability to protect against toxicity from glyphosate and paraquat. The high efficacy of CM and SM shown in this study supports the natural binding activity of glyphosate and paraquat to clay-based soils. Following disasters and medical emergencies, montmorillonite clays could be administered by capsules and tablets, or added to food and flavored water, to reduce toxin bioavailability and human and animal exposures.

Prevalence, serotypes, and drug resistance of nontyphoidal Salmonella among paediatric patients in a tertiary hospital in Guangzhou, China, 2014-2016.

PURPOSE: Nontyphoidal Salmonella (NTS) is a common pathogen responsible for acute gastroenteritis among all ages; however, information on the prevalence, serotypes, and antibiotic susceptibility of NTS isolates is limited. We aimed to explore the characteristics of NTS isolated from paediatric patients in Guangzhou, China. METHODS: This was a retrospective study of 4586 stool culture collected at Guangzhou Women and Children's Medical Center from 2014 to 2016. RESULTS: We identified 220 (4.80%) NTS isolates in stool samples. Fourteen serotypes were identified among the 220 NTS isolates. Salmonella serotype Typhimurium was the most common serotype, representing 69.09%. The highest rate of resistance was recorded in relation to AMP (76.61%), followed by SXT (29.95%), CTX (29.93%), CHL (29.77%), CAZ (23.20%), CIP (7.51%), and CFS (7.18%). The resistance rates of NTS and serotype Typhimurium to CAZ in 2015 were significantly higher than those in 2014. The average hospitalisation duration of inpatients infected by NTS resistant to three or more clinically important agents was significantly longer than that of patients infected with NTS with less antibiotic resistance. CONCLUSION: NTS represents a major cause of paediatric gastroenteritis in Guangzhou, China, and the high level of resistance to third-generation cephalosporins coupled with increasing resistance to quinolones among isolated NTS from paediatric gastroenteritis is a serious public health concern that requires continued monitoring and rational usage of antibiotics.

Seasonal variation in diet and nutrition of the northern-most population of Rhinopithecus roxellana.

There is a great deal of spatial and temporal variation in the availability and nutritional quality of foods eaten by animals, particularly in temperate regions where winter brings lengthy periods of leaf and fruit scarcity. We analyzed the availability, dietary composition, and macronutrients of the foods eaten by the northern-most golden snub-nosed monkey (Rhinopithecus roxellana) population in the Qinling Mountains, China to understand food choice in a highly seasonal environment dominated by deciduous trees. During the warm months between April and November, leaves are consumed in proportion to their availability, while during the leaf-scarce months between December and March, bark and leaf/flower buds comprise most of their diet. When leaves dominated their diet, golden snub-nosed monkeys preferentially selected leaves with higher ratios of crude protein to acid detergent fiber. While when leaves were less available, bark and leaf/flower buds that were high in nonstructural carbohydrates and energy, and low in acid detergent fiber were selected. Southern populations of golden snub-nosed monkey can turn to eating lichen, however, the population studied here in this lichen-absent area have adapted to their cool deciduous habitat by instead consuming buds and bark. Carbohydrate and energy rich foods appear to be the critical resources required for the persistence of this species in temperate habitat. The dietary flexibility of these monkeys, both among seasons and populations, likely contributes to their wide distribution over a range of habitats and environments.

[An effective method of isolation and primary culture of human umbilical arterial endothelial cells].

OBJECTIVE: To explore and establish an efficient system for isolation and culture of human umbilical arterial endothelial cells (HUAECs). METHODS: The human umbilical arteries were perfused and washed by PBS, and then digested by 0.1 g/L type 1 collagenase. Digested cell supernatants were collected, centrifuged, and the re-suspended cells were cultured in a specific medium. The identification of endothelial cells was carried out by observing the cellular morphology, immunostaining the endothelial surface marker CD31 under fluorescent microscopy and examining the capability to form tube-like structure. RESULTS: The isolated cells grew in a monolayer and exhibited mosaic-like pattern under an inverted phase-contrast microscope. They strongly expressed CD31, an endothelium-specific surface marker, and had an obvious capability to form tube-like structure. CONCLUSION: The method established for isolation of HUAECs in this experiment is simple, rapid and efficient. In addition, the isolated cells are of high purity and high survival.

In situ electrochemical assay of cell surface sialic acids featuring highly efficient chemoselective recognition and a dual-functionalized nanohorn probe.

A novel electrochemical strategy was developed for in situ detection and dynamic monitoring of cell surface sialic acids using a highly efficient chemoselective labeling technique and a dual-functionalized nanohorn probe for signal amplification.