MG-BERT: leveraging unsupervised atomic representation learning for molecular property prediction.

MOTIVATION: Accurate and efficient prediction of molecular properties is one of the fundamental issues in drug design and discovery pipelines. Traditional feature engineering-based approaches require extensive expertise in the feature design and selection process. With the development of artificial intelligence (AI) technologies, data-driven methods exhibit unparalleled advantages over the feature engineering-based methods in various domains. Nevertheless, when applied to molecular property prediction, AI models usually suffer from the scarcity of labeled data and show poor generalization ability. RESULTS: In this study, we proposed molecular graph BERT (MG-BERT), which integrates the local message passing mechanism of graph neural networks (GNNs) into the powerful BERT model to facilitate learning from molecular graphs. Furthermore, an effective self-supervised learning strategy named masked atoms prediction was proposed to pretrain the MG-BERT model on a large amount of unlabeled data to mine context information in molecules. We found the MG-BERT model can generate context-sensitive atomic representations after pretraining and transfer the learned knowledge to the prediction of a variety of molecular properties. The experimental results show that the pretrained MG-BERT model with a little extra fine-tuning can consistently outperform the state-of-the-art methods on all 11 ADMET datasets. Moreover, the MG-BERT model leverages attention mechanisms to focus on atomic features essential to the target property, providing excellent interpretability for the trained model. The MG-BERT model does not require any hand-crafted feature as input and is more reliable due to its excellent interpretability, providing a novel framework to develop state-of-the-art models for a wide range of drug discovery tasks.

Influence of optical forces on nonlinear optical frequency conversion in nanoscale waveguide devices.

We theoretically investigate the influence of optical gradient forces on nonlinear frequency conversion in a typical nanoscale optomechanical system, which consists of two parallel, suspended waveguides. The waveguides deform with the input power and the phase-matching wavelength changes along the waveguides. Utilizing the spread of deformations collectively allows phase matching over a wider range of pump wavelengths. The third harmonic phase-matching wavelength shift can be as large as 3.6 nm/mW when the waveguide length is 100 µm and the initial gap is 150 nm. It is analogous to chirping the poling period of quasi-phase-matched devices to extend their bandwidths, and allows broad third harmonic to be generated for uses such as biological spectroscopy. Finally, we discuss the conversion efficiency and the optimal phase-matching wavelength with a single-frequency pump.

Immuno-PCR for detection of Giardia lamblia cysts in water.

Giardia lamblia cysts at low concentrations were detected in water samples using a highly sensitive immunological-PCR (IPCR) method. Magnetic gold particles were precoated with monoclonal anti-Giardia antibodies, and Giardia lamblia cysts ranging from 1 to 100 cysts were diluted in 500 microL of water. A biotinylated detection antibody bound to the G. lamblia cysts was then linked by a streptavidin bridge to biotinylated Giardia-reporter DNA. After extensive washing, reporter DNA was released by denaturation, transferred to PCR tubes, amplified, electrophoresed, and visualized. An optimized immuno-PCR method detected as little as five G. lamblia cysts. To further evaluate the specificity and the clinical application of this IPCR assay for G. lamblia cysts, Entamoeba histolytica and Cryptosporidium parvum were also collected and detected by IPCR. The data demonstrated that this monoclonal antibody-based IPCR method is particularly useful for analysis of environmental water samples in which the densities of G. lamblia cysts is very low, and provides a platform capable of rapid screening of samples from drinking water, wells, rivers, lakes, and recreational swimming pools for trace levels of G. lamblia cysts.

Nicosulfuron stress on the glyoxalase system and endogenous hormone content in sweet maize seedlings.

To reduce the harmful effects of nicosulfuron on sweet corn, the physiological regulation mechanism of sweet corn detoxification was studied. This study analyzed the effects of nicosulfuron stress on the glyoxalase system, hormone content, and key gene expression of nicosulfuron-tolerant "HK301" and nicosulfuron-sensitive "HK320" sweet corn seedling sister lines. After spraying nicosulfuron, the methylglyoxal (MG) content in HK301 increased first and then decreased. Glyoxalase I (GlyI) and glyoxalase II (GlyII) activities, non-enzymatic glutathione (GSH), and the glutathione redox state glutathione/(glutathione + glutathione disulfide) (GSH/(GSH + GSSG)) showed a similar trend as the MG content. Abscisic acid (ABA), gibberellin (GA), and zeatin nucleoside (ZR) also increased first and then decreased, whereas the auxin (IAA) increased continuously. In HK301, all indices after spraying nicosulfuron were significantly greater than those of the control. In HK320, MG accumulation continued to increase after nicosulfuron spraying and GlyI and GlyII activities, and GSH first increased and then decreased after 1 day of stress. The indicators above were significantly greater than the control. The GSH/(GSH + GSSG) ratio showed a decreasing trend and was significantly smaller than the control. Furthermore, ABA and IAA continued to increase, and the GA and ZR first increased and then decreased. Compared with HK320, HK301 significantly upregulated the transcription levels of GlyI and GlyII genes in roots, stems, and leaves. Comprehensive analysis showed that sweet maize seedlings improved their herbicide resistance by changing the glyoxalase system and regulating endogenous hormones. The results provide a theoretical basis for further understanding the response mechanism of the glyoxalase system and the regulation characteristics of endogenous hormones in maize under nicosulfuron stress.

Responses of photosynthesis, antioxidant enzymes, and related gene expression to nicosulfuron stress in sweet maize (Zea mays L.).

Weed control in maize (Zea mays L.) crops is usually undertaken using the postemergence herbicide nicosulfuron. The toxicity of nicosulfuron on maize, especially sweet maize, has been widely reported. In order to examine the effect of nicosulfuron on seedling photosynthetic characteristics, chlorophyll fluorescence, reactive oxygen species production, antioxidant enzyme activities, and gene expressions on sweet maize, nicosulfuron-tolerant "HK310" and nicosulfuron-sensitive "HK320" were studied. All experiment samples were subjected to a water or 80 mg kg-1 of nicosulfuron treatment when sweet maize seedlings grow to the stage of four leaves. After treatment with nicosulfuron, results for HK301 were significantly higher than those for HK320 for net photosynthetic rate, transpiration rate, stomatal conductance, leaf maximum photochemical efficiency of PSII, photochemical quenching of chlorophyll fluorescence, and the electron transport rate. These results were contrary to nonphotochemical quenching and intercellular CO2 concentration. As exposure time increased, associated effects also increased. Both O2·- and H2O2 detoxification is modulated by antioxidant enzymes. Compared to HK301, SOD, POD, and CAT activities of HK320 were significantly reduced as exposure time increase. Compared to HK320, the gene expression for the majority of SOD genes, except for SOD2, increased due to inducement by nicosulfuron, and it significantly upregulated the gene expression of CAT in HK301. Results from this study indicate that plants can improve photosynthesis, scavenging capabilities of ROS, and protective mechanisms to alleviate phytotoxic effect of nicosulfuron. Future research is needed to further elucidate the important role antioxidant systems and gene regulation play in herbicide detoxification in sweet maize.

Molecular diversity and relationships among Cymbidium goeringii cultivars based on inter-simple sequence repeat (ISSR) markers.

Spring orchid (Cymbidium goeringii) is a popular flowering plant species. There have been few molecular studies of the genetic diversity and conservation genetics on this species. An assessment of the level of genetic diversity in cultivated spring orchid would facilitate development of the future germplasm conservation for cultivar improvement. In the present study, DNA markers of intersimple sequence repeats (ISSR) were identified and the ISSR fingerprinting technique was used to evaluate genetic diversity in C. goeringii cultivars. Twenty-five ISSR primers were selected to produce a total of 224 ISSR loci for evaluation of the genetic diversity. A wide genetic variation was found in the 50 tested cultivars with Nei's gene diversity (H = 0.2241) and 93.75% of polymorphic loci. Fifty cultivars were unequivocally distinguished based on ISSR fingerprinting. Cultivar-specific ISSR markers were identified in seven of 50 tested cultivars. Unweighted pair-group mean analysis (UPGMA) and principal coordinates analysis (PCA) grouped them into two clusters: one composed the cultivars mainly from Japan, and the other contained three major subclusters mainly from China. Two Chinese subclusters were generally consistent with horticultural classification, and the third Chinese subcluster contained cultivars from various horticultural groups. Our results suggest that the ISSR technique provides a powerful tool for cultivar identification and establishment of genetic relationships of cultivars in C. goeringii.

[The genetic diversity of Cymbidium by ISSR].

ISSR was applied to detect the relationship between 16 Cymbidium species, and 836 bands were amplified with 15 primers, including 227 polymorphic bands. The polymorphic percentage is 27.2%. UPGMA results showed that the genetic distance were closest between C.goeringii (Rchb.f.) Rchb.f. and C. goeringii var. longibracteatum, and C.lancifolium Hook. was far away from the other 15 species. This result is quite similar to the traditional classification, indicating that the technique could supplement some information to traditional taxonomy in the molecular level.

Apoptosis induced by yessotoxins in Hela human cervical cancer cells in vitro.

To investigate the apoptotic effects of an extract of red-tide algae, Protoceratium reticulatum, on HeLa cells, we used liquid chromatography-tandem mass spectrometry, optical microscopy, Hoechst 33342/propidine iodide staining, and DNA gel electrophoresis to analyze its constituents and toxicity, as well as rhodamine 123 staining to investigate changes in mitochondrial membrane potential. Analysis showed that the P. reticulatum extract contained the yessotoxins (YTXs) homo-YTX, 45-OH-YTX and 45-OH-homo-YTX. The results indicated that P. reticulatum extract negatively influences HeLa cells and induces their apoptosis.

[Study on the relationships of Cymbidium by RMAPD technique].

Cymbidium is one of the Orchidaceae genera. There are 70 Cymbidium species in the world, mainly distributed over the tropical and the subtropical area of Asia and northern Australia. In China, approximately 49 species and some variations are distributed over the provinces to the south of Qinling. The orchid market of China is quite wide, not only because of their view and admiration value, but also their officinal value as well. Some species of Cymbidium are morphologically similar, so the identification of Cymbidium species and the understanding of the relationships between them are very important for new varieties breeding of Cymbidium. This study attempts to apply RMAPD to detect the relationships between the Cymbidium species. 456 pairs of RMAPD combinations were screened, but only 12 pairs of them were used to analyze the genetic polymorphism of 16 Cymbidium. 362 bands were produced by 12 pairs of RMAPD primers in total, including 90 polymorphic bands (24.8%). The polymorphic bands of each primer combination ranged from 3 to 11 with the average of 8. Among all species, genetic similarity coefficient ranged from 0.63 to 0.87. The result showed that the genetic distance was closest between C. floribundum and C. aestivum, C. ensifolium and C. sinense, C. maguanense and C. hookerianum, which is almost the same as the traditional classification. It was the first time that species of Cymbidium were detected by RMAPD technique. This study shows RMAPD is an appropriate and effective molecular marker technique to identify the relationships of Cymbidium.