GCMSFormer: A Fully Automatic Method for the Resolution of Overlapping Peaks in Gas Chromatography-Mass Spectrometry.

Gas chromatography-mass spectrometry (GC-MS) is one of the most important instruments for analyzing volatile organic compounds. However, the complexity of real samples and the limitations of chromatographic separation capabilities lead to coeluting compounds without ideal separation. In this study, a Transformer-based automatic resolution method (GCMSFormer) is proposed to resolve mass spectra from GC-MS peaks in an end-to-end manner, predicting the mass spectra of components directly from the raw overlapping peaks data. Furthermore, orthogonal projection resolution (OPR) was integrated into GCMSFormer to resolve minor components. The GCMSFormer model was trained, validated, and tested using 100,000 augmented data. It achieves 99.88% of the bilingual evaluation understudy (BLEU) value on the test set, significantly higher than the 97.68% BLEU value of the baseline sequence-to-sequence model long short-term memory (LSTM). GCMSFormer was also compared with two nondeep learning resolution tools (MZmine and AMDIS) and two deep learning resolution tools (PARAFAC2 with DL and MSHub/GNPS) on a real plant essential oil GC-MS data set. Their resolution results were compared on evaluation metrics, including the number of compounds resolved, mass spectral match score, correlation coefficient, explained variance, and resolution speed. The results demonstrate that GCMSFormer has better resolution performance, higher automation, and faster resolution speed. In summary, GCMSFormer is an end-to-end, fast, fully automatic, and accurate method for analyzing GC-MS data of complex samples.

Lewis Acidic Metal-Organic Framework Assisted Ambient Liquid Extraction Mass Spectrometry Imaging for Enhancing the Coverage of Poorly Ionizable Lipids in Brain Tissue.

The spatial distribution of lipidomes in tissues is of great importance in studies of living processes, diseases, and therapies. Mass spectrometry imaging (MSI) has become a critical technique for spatial lipidomics. However, MSI of low-abundance or poorly ionizable lipids is still challenging because of the ion suppression from high-abundance lipids. Here, a metal-organic framework (MOF) Zr6O4(OH)4(1,3,5-Tris(4-carboxyphenyl) benzene)2(triflate)6(Zr6OTf-BTB) was prepared and used for selective on-tissue adsorption of phospholipids to reduce ion suppression from them to poorly ionizable lipids. The results show that Zr6OTf-BTB with strong Lewis acidic sites and a large specific surface area (647.9 m2·g-1) could selectively adsorb phospholipids under 1% FA-MeOH. Adsorption efficiencies of phospholipids are 88.4-144.9 times higher than those of other neutral lipids. Moreover, the adsorption capacity and the adsorption kinetic rate constant of the new material to phospholipids are higher than those of Zr6-BTB (242.72 vs 73.96 mg·g-1, 0.0442 vs 0.0220 g·mg-1·min-1). A Zr6OTf-BTB sheet was prepared by a lamination technique for on-tissue phospholipid adsorption from brain tissue. Then, the tissue section on the Zr6OTf-BTB sheet was directly imaged via ambient liquid extraction-MSI with 1% FA-MeOH as the sampling solvent. The results showed that phospholipids could be 100% removed directly on tissue, and the detection coverage of the Zr6OTf-BTB-enhanced MSI method to ceramides (Cers) and hexosylceramides (HexCers) was increased by 5-26 times compared with direct tissue MSI (26 vs 1 and 17 vs 3). The new method provides an efficient and convenient way to eliminate the ion suppression from phospholipids in MSI, largely improving the detection coverage of low-abundance and poorly ionizable lipids.

Correction: Sperm associated antigen 9 promotes oncogenic KSHV-encoded interferon regulatory factor-induced cellular transformation and angiogenesis by activating the JNK/VEGFA pathway.

[This corrects the article DOI: 10.1371/journal.ppat.1008730.].

A novel lncRNA TCONS_00071187 upregulated by activated GSK3ß promotes high glucose-induced mesangial cell proliferation in the diabetic nephropathy.

Inhibiting mesangial cell proliferation is one of the strategies to control the early progression of diabetic nephropathy (DN). GSK3ß is closely related to cell apoptosis as well as the development of DN, but whether it acts on the proliferation of mesangial cells is unclear. This study aimed to elucidate the role and mechanism of GSK3ß-mediated lncRNA in high glucose-induced mesangial cell proliferation. HBZY-1 cells were used to establish the cell model of DN. The automatic cell counter was applied to assess cell proliferation. Flow cytometry was used to detect cell apoptosis and intracellular ROS levels. High-throughput transcriptomics sequencing was performed to detect the different expressions of long noncoding RNAs (lncRNAs) in the cell model of DN after knocking down the expression of GSK3ß by the transfection of siRNA. The expression of RNA was detected by real-time PCR. In the cell model of DN using HBZY-1 cells, cell proliferation was enhanced accompanied by GSK3ß activation and elevated apoptosis rate and reactive oxygen species (ROS) levels. A panel of novel lncRNAs, which were differentially expressed after GSK3ß knockdown in the cell model of DN, were identified by high-throughput transcriptomics sequencing. Among them, the expression of TCONS_00071187 was upregulated under high glucose conditions while the knockdown of the GSK3ß expression led to the downregulation of TCONS_00071187. The knockdown of TCONS_00071187 resulted in reduced mesangial cell proliferation, and decreased apoptosis rates and ROS levels. In conclusion, GSK3ß promoted mesangial cell proliferation by upregulating TCONS_00071187, which led to enhanced ROS production under high glucose conditions in the cell model of DN. This study revealed the role of GSK3ß medicated lncRNAs in the development of DN.

The influence of resourcefulness on the family functioning of young- and middle-aged lymphoma patients in China: a cross-sectional study.

BACKGROUND: Lymphoma has become 1 of the 10 most common cancers with increased prevalence in young- and middle-aged adults in China. This poses a tremendous burden on patients and their families and brings great challenges to maintaining the balance of family functioning in young- and middle-aged patients. OBJECTIVE: This cross-sectional study aimed to analyse the influence of resourcefulness on the family functioning of Chinese young- and middle-aged lymphoma patients. METHODS: A total of 172 Chinese young- and middle-aged patients with lymphoma were recruited from the oncology departments of two tertiary hospitals in Zhengzhou, Henan, China. They were invited to complete a survey that included a demographic questionnaire, the Resourcefulness Scale and the Chinese Version Family Adaptability and Cohesion Scale II. Multiple linear regression was used to analyse the related factors for family functioning. RESULTS: The multiple regression analysis revealed that the main influencing factors of family cohesion were resourcefulness (ß = 0.338, 95% CI (0.072, 0.173)), spouse caregiver (ß = 0.376, 95% CI (1.938, 10.395)), and cancer stage (ß = -0.274, 95% CI (-3.219, -1.047)). Resourcefulness (ß = 0.438, 95% CI (0.096, 0.181)), spouse caregiver (ß = 0.340, 95% CI (1.348, 8.363)), and family per capita monthly income (ß = 0.157, 95% CI (0.066, 2.243)) were the influencing factors of family adaptability. CONCLUSIONS: Healthcare professionals and family scholars should value young- and middle-aged lymphoma patients' family functioning throughout the cancer treatment process, and family interventions should be designed by healthcare providers based on patients' resourcefulness. Moreover, healthcare providers need to pay attention to the risk factors of patients' family cohesion and adaptability, such as low family per capita monthly income, and consider employing corresponding measures to help them.

Pulled Flowprobe for Ambient Liquid Extraction-Based High Spatial Resolution Mass Spectrometry Imaging with Enhanced Sensitivity and Stability.

Ambient liquid extraction techniques enable direct mass spectrometry imaging (MSI) under ambient conditions with minimal sample preparation. However, currently an integrated probe for ambient liquid extraction-based MSI with high spatial resolution, high sensitivity, and stability is still lacking. In this work, we developed a new integrated probe made of pulled coaxial capillaries, named pulled flowprobe, and compared it with the previously reported single-probe. Mass transfer kinetics in probes was first investigated. The extraction kinetic curves during probe sampling indicate a narrower and higher peak shape for the pulled flowprobe than single-probe. Computational fluid dynamics analysis reveals that in the pulled flowprobe flow velocities are lower in liquid microjunction and higher in the transferring channels, resulting in higher extraction efficiencies and reduced band diffusion compared with single-probe and other probes with a similar flow route. Results of ambient liquid extraction-based MSI of lipids in rat cerebrum show that signals of low-abundance lipids were 2-5 times higher via a pulled flowprobe than via a single-probe, and 26 more lipid species were detected on brain tissue via a pulled flowprobe than via a single-probe. The stability of MSI with the pulled flowprobe was found to be higher than that with single-probe (averaged relative standard deviation = 18% vs 80%) by imaging a lab-made uniform ink coating. Moreover, in the pulled flowprobe, no retraction of the inner capillary from outer capillary is optimal for both sensitivity and stability. The spatial resolution of the pulled flowprobe (30-40 µm) was measured to be higher than that of a comparable size single-probe by calculation with the "80-20" rule. Finally, the new pulled flowprobe was applied to high-resolution MSI of lipids in the hippocampus, and localization of several lipids to the specific cell layers in the hippocampus region was observed. Thus, this work provides an alternative easily fabricated sampling probe with enhanced sensitivity, stability, and spatial resolution, promoting the use of ambient liquid extraction-based MSI in biological and clinical research.

A Universal and Accurate Method for Easily Identifying Components in Raman Spectroscopy Based on Deep Learning.

Raman spectroscopy has been widely used to provide the structural fingerprint for molecular identification. Due to interference from coexisting components, noise, baseline, and systematic differences between spectrometers, component identification with Raman spectra is challenging, especially for mixtures. In this study, a method entitled DeepRaman has been proposed to solve those problems by combining the comparison ability of a pseudo-Siamese neural network (pSNN) and the input-shape flexibility of spatial pyramid pooling (SPP). DeepRaman was trained, validated, and tested with 41,564 augmented Raman spectra from two databases (pharmaceutical material and S.T. Japan). It can achieve 96.29% accuracy, 98.40% true positive rate (TPR), and 94.36% true negative rate (TNR) on the test set. Another six data sets measured on different instruments were used to evaluate the performance of the proposed method from different aspects. DeepRaman can provide accurate identification results and significantly outperform the hit quality index (HQI) method and other deep learning models. In addition, it performs well in cases of different spectral complexity and low-content components. Once the model is established, it can be used directly on different data sets without retraining or transfer learning. Furthermore, it also obtains promising results for the analysis of surface-enhanced Raman spectroscopy (SERS) data sets and Raman imaging data sets. In summary, it is an accurate, universal, and ready-to-use method for component identification in various application scenarios.

Fusion of Quality Evaluation Metrics and Convolutional Neural Network Representations for ROI Filtering in LC-MS.

Region of interest (ROI) extraction is a fundamental step in analyzing metabolomic datasets acquired by liquid chromatography-mass spectrometry (LC-MS). However, noises and backgrounds in LC-MS data often affect the quality of extracted ROIs. Therefore, developing effective ROI evaluation algorithms is necessary to eliminate false positives meanwhile keep the false-negative rate as low as possible. In this study, a deep fused filter of ROIs (dffROI) was proposed to improve the accuracy of ROI extraction by combining the handcrafted evaluation metrics with convolutional neural network (CNN)-learned representations. To evaluate the performance of dffROI, dffROI was compared with peakonly (CNN-learned representation) and five handcrafted metrics on three LC-MS datasets and a gas chromatography-mass spectrometry (GC-MS) dataset. Results show that dffROI can achieve higher accuracy, better true-positive rate, and lower false-positive rate. Its accuracy, true-positive rate, and false-positive rate are 0.9841, 0.9869, and 0.0186 on the test set, respectively. The classification error rate of dffROI (1.59%) is significantly reduced compared with peakonly (2.73%). The model-agnostic feature importance demonstrates the necessity of fusing handcrafted evaluation metrics with the convolutional neural network representations. dffROI is an automatic, robust, and universal method for ROI filtering by virtue of information fusion and end-to-end learning. It is implemented in Python programming language and open-sourced at https://github.com/zhanghailiangcsu/dffROI under BSD License. Furthermore, it has been integrated into the KPIC2 framework previously proposed by our group to facilitate real metabolomic LC-MS dataset analysis.

Prediction of drug-likeness using graph convolutional attention network.

MOTIVATION: The drug-likeness has been widely used as a criterion to distinguish drug-like molecules from non-drugs. Developing reliable computational methods to predict the drug-likeness of compounds is crucial to triage unpromising molecules and accelerate the drug discovery process. RESULTS: In this study, a deep learning method was developed to predict the drug-likeness based on the graph convolutional attention network (D-GCAN) directly from molecular structures. Results showed that the D-GCAN model outperformed other state-of-the-art models for drug-likeness prediction. The combination of graph convolution and attention mechanism made an important contribution to the performance of the model. Specifically, the application of the attention mechanism improved accuracy by 4.0%. The utilization of graph convolution improved the accuracy by 6.1%. Results on the dataset beyond Lipinski's rule of five space and the non-US dataset showed that the model had good versatility. Then, the billion-scale GDB-13 database was used as a case study to screen SARS-CoV-2 3C-like protease inhibitors. Sixty-five drug candidates were screened out, most substructures of which are similar to these of existing oral drugs. Candidates screened from S-GDB13 have higher similarity to existing drugs and better molecular docking performance than those from the rest of GDB-13. The screening speed on S-GDB13 is significantly faster than screening directly on GDB-13. In general, D-GCAN is a promising tool to predict the drug-likeness for selecting potential candidates and accelerating drug discovery by excluding unpromising candidates and avoiding unnecessary biological and clinical testing. AVAILABILITY AND IMPLEMENTATION: The source code, model and tutorials are available at https://github.com/JinYSun/D-GCAN. The S-GDB13 database is available at https://doi.org/10.5281/zenodo.7054367. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Establishment of a model for predicting preterm birth based on the machine learning algorithm.

BACKGROUND: The purpose of this study was to construct a preterm birth prediction model based on electronic health records and to provide a reference for preterm birth prediction in the future. METHODS: This was a cross-sectional design. The risk factors for the outcomes of preterm birth were assessed by multifactor logistic regression analysis. In this study, a logical regression model, decision tree, Naive Bayes, support vector machine, and AdaBoost are used to construct the prediction model. Accuracy, recall, precision, F1 value, and receiver operating characteristic curve, were used to evaluate the prediction performance of the model, and the clinical application of the model was verified. RESULTS: A total of 5411 participants were included and were used for model construction. AdaBoost model has the best prediction ability among the five models. The accuracy of the model for the prediction of "non-preterm birth" was the highest, reaching 100%, and that of "preterm birth" was 72.73%. CONCLUSIONS: By constructing a preterm birth prediction model based on electronic health records, we believe that machine algorithms have great potential for preterm birth identification. However, more relevant studies are needed before its application in the clinic.

Deep-Learning-Based Mixture Identification for Nuclear Magnetic Resonance Spectroscopy Applied to Plant Flavors.

Nuclear magnetic resonance (NMR) is a crucial technique for analyzing mixtures consisting of small molecules, providing non-destructive, fast, reproducible, and unbiased benefits. However, it is challenging to perform mixture identification because of the offset of chemical shifts and peak overlaps that often exist in mixtures such as plant flavors. Here, we propose a deep-learning-based mixture identification method (DeepMID) that can be used to identify plant flavors (mixtures) in a formulated flavor (mixture consisting of several plant flavors) without the need to know the specific components in the plant flavors. A pseudo-Siamese convolutional neural network (pSCNN) and a spatial pyramid pooling (SPP) layer were used to solve the problems due to their high accuracy and robustness. The DeepMID model is trained, validated, and tested on an augmented data set containing 50,000 pairs of formulated and plant flavors. We demonstrate that DeepMID can achieve excellent prediction results in the augmented test set: ACC = 99.58%, TPR = 99.48%, FPR = 0.32%; and two experimentally obtained data sets: one shows ACC = 97.60%, TPR = 92.81%, FPR = 0.78% and the other shows ACC = 92.31%, TPR = 80.00%, FPR = 0.00%. In conclusion, DeepMID is a reliable method for identifying plant flavors in formulated flavors based on NMR spectroscopy, which can assist researchers in accelerating the design of flavor formulations.

Porous Graphitic Carbon-Based Imprint Mass Spectrometry Imaging with an Ambient Liquid Extraction Technique for Enhancing Coverage of Glycerolipids and Sphingolipids in Brain Tissue.

Localization of lipidomes and tracking their spatial changes by mass spectrometry imaging (MSI) is critical for the mechanism studies on living process, disease, and therapeutic treatment. However, due to the strong ion suppression in complex biotissue, the imaging coverage for lipids with low polarity or low abundances, such as glycerolipids and sphingolipids, is usually limited. To address this issue, we utilized a porous graphitic carbon (PGC) material to imprint brain tissue sections for selective enrichment of neutral lipids with polar phospholipids removed. Then, the tissue imprint was scanned for desorption by the ambient liquid extraction MSI system. It was found that on the PGC surface, hydrophobic interaction dominates in protic solvents, and polar interaction dominates in aprotic solvents. Accordingly, methanol was selected as the spray solvent for tissue imprinting, and 75% acetonitrile-methanol was selected as the desorption solvent for the ambient liquid extraction MSI system. The results showed that glycerides had high recoveries after the imprinting-desorption process (recovery ∼ 70%) with phospholipids eliminated (recovery < 7%). To increase the transferring efficiencies of lipids from tissue onto PGC, electrospray was used for solvent application during imprinting, and the signals of diglycerides (DGs) in the imprint MSI of brain tissue increased by 2-3 times as compared to those via air spray. Finally, the new imprint MSI approach was applied to the imaging of the rat cerebellum and was compared with direct tissue MSI. The results showed that with imprint MSI, the coverage of DGs, sphingomyelins (SMs), and ceramides was enhanced by 4-5-fold (32 vs 6, 4 vs 1, and 5 vs 0). The ion images showed that with imprint MSI, higher signal intensities and clearer spatial distribution of DGs and SMs were obtained without interference from phosphatidylcholine signals compared with tissue MSI. This new method provides a complementary approach for traditional MSI to address the issues in imaging poorly ionizable or low-abundance lipids.

Selective Capture-Based Single-Cell Mass Spectrometry for Enhancing Sphingolipid Profiling of Neurons with Differentiation of Cell Body from Synapse.

Currently, single-cell lipidomic mass spectrometry (MS) techniques are mostly limited to detection of high-abundance phosphatidylcholines (PCs). Herein, for enhancing the coverage to low-abundance sphingolipids in single-cell analysis, in-tube solid-phase microextraction (SPME) was combined with a single-probe MS system for selective enrichment of sphingolipids during singe-cell sampling. From the results, a lab-made single probe with a 30 µm tip size proved to be able to resolve the axon from the cell body of neuron HT22 in ambient conditions. TiO2 was immobilized onto the inner wall of the transfer capillary of the single-probe device for online selective capture of sphingolipids in ammonia-acetonitrile and rapid desorption in formic acid-methanol. The results showed that the breakthrough volume of the capillary with sample loading flow rate at 500 nL/min was >14 µL. Standard experiments showed that the signals of cerebroside (CB), ceramide (Cer), and sphingomyelin (SM) were largely enhanced after selective capture in the coated capillary, while PCs were totally removed. The reusability (>10 times) and stability of the lab-made TiO2-coated capillary was verified. By introducing the coated capillary into the single-probe MS system, the new system proved to have low detection limits of SM, Cer, and CB (0.007-0.027 µg/mm2) and acceptable linearity (r > 0.98) and repeatability (RSD < 30%). Lipid coverage of the new method to SMs and CBs proved to be largely improved (SM, 21 vs 2; CB, 10 vs 0) with the new method in comparison to conventional single-probe MS without selective capture by ambient analysis of a single spot of rat cerebellum. Finally, the new system was used to perform single-neuron analysis of sphingolipids in the control and lipopolysaccharide (LPS)-treated HT22 with differentiation of the cell body from the axonal synapse. Results showed that 5 sphingolipids had significantly higher concentrations in the synapse than in the cell body, while 3 oxidized sphingolipids had significantly higher levels in the cell body than in the synapse. After LPS treatment, most of the sphingolipids largely decreased and became more accumulated in the synapse, providing new information on LPS-induced neuroinflammation.

Sperm associated antigen 9 promotes oncogenic KSHV-encoded interferon regulatory factor-induced cellular transformation and angiogenesis by activating the JNK/VEGFA pathway.

Kaposi's sarcoma (KS), caused by Kaposi's sarcoma-associated herpesvirus (KSHV), is a highly angioproliferative disseminated tumor of endothelial cells commonly found in AIDS patients. We have recently shown that KSHV-encoded viral interferon regulatory factor 1 (vIRF1) mediates KSHV-induced cell motility (PLoS Pathog. 2019 Jan 30;15(1):e1007578). However, the role of vIRF1 in KSHV-induced cellular transformation and angiogenesis remains unknown. Here, we show that vIRF1 promotes angiogenesis by upregulating sperm associated antigen 9 (SPAG9) using two in vivo angiogenesis models including the chick chorioallantoic membrane assay (CAM) and the matrigel plug angiogenesis assay in mice. Mechanistically, vIRF1 interacts with transcription factor Lef1 to promote SPAG9 transcription. vIRF1-induced SPAG9 promotes the interaction of mitogen-activated protein kinase kinase 4 (MKK4) with JNK1/2 to increase their phosphorylation, resulting in enhanced VEGFA expression, angiogenesis, cell proliferation and migration. Finally, genetic deletion of ORF-K9 from KSHV genome abolishes KSHV-induced cellular transformation and impairs angiogenesis. Our results reveal that vIRF1 transcriptionally activates SPAG9 expression to promote angiogenesis and tumorigenesis via activating JNK/VEGFA signaling. These novel findings define the mechanism of KSHV induction of the SPAG9/JNK/VEGFA pathway and establish the scientific basis for targeting this pathway for treating KSHV-associated cancers.

Discovery of New Secondary Metabolites from Marine Bacteria Hahella Based on an Omics Strategy.

Hahella is one characteristic genus under the Hahellaceae family and shows a good potential for synthesizing new natural products. In this study, we examined the distribution of the secondary metabolite biosynthetic gene cluster (SMBGC) under Hahella with anti-SMASH. The results derived from five genomes released 70 SMBGCs. On average, each strain contains 12 gene clusters, and the most abundant ones (45.7%) are from the family of non-ribosomal peptide synthetase (NRPS) and non-ribosomal peptide synthetase hybrid with polyketide synthase (NRPS/PKS), indicating a great potential to find bioactive compounds. The comparison of SMBGC between H. chejuensis and other species showed that H. chejuensis contained two times more gene clusters than H. ganghwensis. One strain, designed as NBU794, was isolated from the mangrove soil of Dongzhai Port in Haikou (China) by iChip. The 16S rRNA gene of NBU794 exhibited 99% identity to H. chejuensis KCTC 2396 and clustered with the H. chejuensis clade on the phylogenetic trees. Genome mining on strain NBU794 released 17 SMBGCs and two groups of bioactive compounds, which are chejuenolide A-C and nine prodiginines derivatives. The prodiginines derivatives include the well-known lead compound prodigiosin and two new compounds, 2-methyl-3-pentyl-4-O-methyl-prodiginine and 2-methyl-3-octyl-prodiginine, which were identified through fragmentation analysis based on LC-MS/MS. The anti-microbial activity assay showed prodigiosin and 2-methyl-3-heptyl-prodiginine exhibited the best performance in inhibiting Escherichia coli, Salmonella paratyphi B, MASA Staphylococcus aureus, Bacillus subtilis, and Candida albicans. Moreover, the yield of prodigiosin in H. chejuensis NBU794 was also evaluated, which could reach 1.40 g/L under the non-optimized condition and increase to 5.83 g/L in the modified ISP4 medium with macroporous adsorption beads added, indicating that NBU794 is a promising source of prodigiosin.

Deep Learning-Based Method for Compound Identification in NMR Spectra of Mixtures.

Nuclear magnetic resonance (NMR) spectroscopy is highly unbiased and reproducible, which provides us a powerful tool to analyze mixtures consisting of small molecules. However, the compound identification in NMR spectra of mixtures is highly challenging because of chemical shift variations of the same compound in different mixtures and peak overlapping among molecules. Here, we present a pseudo-Siamese convolutional neural network method (pSCNN) to identify compounds in mixtures for NMR spectroscopy. A data augmentation method was implemented for the superposition of several NMR spectra sampled from a spectral database with random noises. The augmented dataset was split and used to train, validate and test the pSCNN model. Two experimental NMR datasets (flavor mixtures and additional flavor mixture) were acquired to benchmark its performance in real applications. The results show that the proposed method can achieve good performances in the augmented test set (ACC = 99.80%, TPR = 99.70% and FPR = 0.10%), the flavor mixtures dataset (ACC = 97.62%, TPR = 96.44% and FPR = 2.29%) and the additional flavor mixture dataset (ACC = 91.67%, TPR = 100.00% and FPR = 10.53%). We have demonstrated that the translational invariance of convolutional neural networks can solve the chemical shift variation problem in NMR spectra. In summary, pSCNN is an off-the-shelf method to identify compounds in mixtures for NMR spectroscopy because of its accuracy in compound identification and robustness to chemical shift variation.

Trends and influencing factors of perinatal birth defects in Huai'an from 2008 to 2020.

OBJECTIVE: To analyze the incidence, trends and related factors of birth defects in Huai'an from 2008 to 2020. METHODS: The surveillance data from maternal and child health system of Huai'an from 2008 to 2020 and Huai'an Statistical Yearbook were used for analysis. Taking the annual change percentage and average annual change percentage (AAPC) as the main outcome indicators, the JoinPoint regression analysis was performed to estimate the changing trend of birth defects from 2008 to 2020. Spearman correlation analysis was used to examine the association between birth defects and birth rate, marriage rate, proportion of women with advanced maternal age. RESULTS: During 2008 to 2020, a total of 3414 cases of neonatal birth defects occurred in Huai'an, with an incidence of 4.6‰ (3414/736 608). The rate of perinatal birth defects in Huai'an showed an increasing trend (AAPC=8.8%, t=3.2, P<0.01), and the year of 2016 was a significant changing point. Among 24 types of birth defects, the incidence of congenital heart disease rose and became the most prevalent defect, while the incidence of neural tube malformations such as anencephaly, encephalocele and spina bifida was declined. The incidence of birth defect was negatively correlated with the birth rate ( r=-0.751, P<0.01), not correlated with marriage rate ( r=-0.516, P>0.05), and positively correlated with the proportion of women with advanced maternal age ( r=0.726, P<0.01). CONCLUSION: The incidence of birth defects in Huai'an shows an increasing trend from 2008 to 2020 with congenital heart disease as the most common type of birth defect, and the increase of birth defects incidence is closely related with the increase of the proportion of women with advanced maternal age.

Prediction of Liquid Chromatographic Retention Time with Graph Neural Networks to Assist in Small Molecule Identification.

The predicted liquid chromatographic retention times (RTs) of small molecules are not accurate enough for wide adoption in structural identification. In this study, we used the graph neural network to predict the retention time (GNN-RT) from structures of small molecules directly without the requirement of molecular descriptors. The predicted accuracy of GNN-RT was compared with random forests (RFs), Bayesian ridge regression, convolutional neural network (CNN), and a deep-learning regression model (DLM) on a METLIN small molecule retention time (SMRT) dataset. GNN-RT achieved the highest predicting accuracy with a mean relative error of 4.9% and a median relative error of 3.2%. Furthermore, the SMRT-trained GNN-RT model can be transferred to the same type of chromatographic systems easily. The predicted RT is valuable for structural identification in complementary to tandem mass spectra and can be used to assist in the identification of compounds. The results indicate that GNN-RT is a promising method to predict the RT for liquid chromatography and improve the accuracy of structural identification for small molecules.

Photocatalytic reduction-based liquid microjunction surface sampling-mass spectrometry for rapid in situ analysis of aromatic amines originating from azo dyes in packaging papers.

A rapid in situ analytical method was developed for the detection of generated carcinogenic aromatic amines from banned azo dyes utilizing a photocatalytic reduction-based liquid microjunction surface sampling (LMJSS)-mass spectrometry (MS) system. We utilized photocatalytic reduction under UV irradiation with TiO2 as catalyst to have rapid and mild reduction of azo dyes. The reaction conditions were optimized to have complete photocatalytic reduction within 2-5 min in pure methanol at room temperature. TiO2 was immobilized in the inner wall of the capillaries in the LMJSS system to achieve in situ sampling-online rapid reduction-MS detection for aromatic amines originating from azo dyes in packaging surface. The yields of in-tube photocatalytic reduction were near 100% by delivering the azo dye extracts through the capillary at 1 µL/min under UV irradiation. With this design, in situ analysis was completed within 2 min via direct MS detection and 7 min via liquid chromatography (LC)-MS detection. The detection limits for five aromatic amines originating from four different azo dyes were in the range of 1-17 mg/kg with relative standard deviations (RSDs) < 8.5%. In the application of the new method, four carcinogenic aromatic amines were detected and identified in three commercial packaging materials, and the quantitation results were comparable with those obtained by the conventional chemical reduction-LC-MS method (relative recovery, 81-121%). Moreover, due to the spatial resolution of the present method with a flow probe, MS imaging was achieved demonstrating clear azo dye patterns of a lab-made sample.

Pure Ion Chromatograms Combined with Advanced Machine Learning Methods Improve Accuracy of Discriminant Models in LC-MS-Based Untargeted Metabolomics.

Untargeted metabolomics based on liquid chromatography coupled with mass spectrometry (LC-MS) can detect thousands of features in samples and produce highly complex datasets. The accurate extraction of meaningful features and the building of discriminant models are two crucial steps in the data analysis pipeline of untargeted metabolomics. In this study, pure ion chromatograms were extracted from a liquor dataset and left-sided colon cancer (LCC) dataset by K-means-clustering-based Pure Ion Chromatogram extraction method version 2.0 (KPIC2). Then, the nonlinear low-dimensional embedding by uniform manifold approximation and projection (UMAP) showed the separation of samples from different groups in reduced dimensions. The discriminant models were established by extreme gradient boosting (XGBoost) based on the features extracted by KPIC2. Results showed that features extracted by KPIC2 achieved 100% classification accuracy on the test sets of the liquor dataset and the LCC dataset, which demonstrated the rationality of the XGBoost model based on KPIC2 compared with the results of XCMS (92% and 96% for liquor and LCC datasets respectively). Finally, XGBoost can achieve better performance than the linear method and traditional nonlinear modeling methods on these datasets. UMAP and XGBoost are integrated into KPIC2 package to extend its performance in complex situations, which are not only able to effectively process nonlinear dataset but also can greatly improve the accuracy of data analysis in non-target metabolomics.