Annotation Propagation Research Articles

Introducing Molecular Hypernetworks for Discovery in Multidimensional Metabolomics Data.

Orthogonal separations of data from high-resolution mass spectrometry can provide insight into sample composition and address challenges of complete annotation of molecules in untargeted metabolomics. "Molecular networks" (MNs), as used in the Global Natural Products Social Molecular Networking platform, are a prominent strategy for exploring and visualizing molecular relationships and improving annotation. MNs are mathematical graphs showing the relationships between measured multidimensional data features. MNs also show promise for using network science algorithms to automatically identify targets for annotation candidates and to dereplicate features associated with a single molecular identity. This paper introduces "molecular hypernetworks" (MHNs) as more complex MN models able to natively represent multiway relationships among observations. Compared to MNs, MHNs can more parsimoniously represent the inherent complexity present among groups of observations, initially supporting improved exploratory data analysis and visualization. MHNs also promise to increase confidence in annotation propagation, for both human and analytical processing. We first illustrate MHNs with simple examples, and build them from liquid chromatography- and ion mobility spectrometry-separated MS data. We then describe a method to construct MHNs directly from existing MNs as their "clique reconstructions", demonstrating their utility by comparing examples of previously published graph-based MNs to their respective MHNs.

Feature-based molecular networking with MS2LDA to profile compounds in Lanbuzheng based on ultra-high-performance liquid chromatography-quadrupole Exactive Orbitrap high-resolution mass spectrometry.

Lanbuzheng (Geum japonicum Thunb. var. chinense Bolle), a plant found in Southwest China, is a traditional Chinese medicine that promotes hematopoiesis and antioxidant functions. Many of its chemical constituents remain unknown, posing challenges both to understanding its pharmacological mechanisms and to conducting quality control research. In this work, ultra-high performance liquid chromatography coupled with quadrupole Exactive Orbitrap high-resolution mass spectroscopy was used for profiling the composition of Lanbuzheng. Using positive ion mass spectrometry data enriched from Lanbuzheng extract, feature-based molecular networking (FBMN) was constructed and associated with Mass2Motifs substructures using MS2LDA. Prediction and validation of unknown constituents of Lanbuzheng using a custom-built compound library, SIRIUS, and network annotation propagation, achieved a semi-automated annotation of the molecular network. Based on the custom-built library comprising 206 compounds and the FBMN clustering results, the constituents in Lanbuzheng primarily include tannins, triterpenes, flavonoids, and phenolics. Using only 65 pre-identified compounds as references, 210 unknown compounds were annotated in various polarity regions of Lanbuzheng. Results of the current work indicate that molecular networks enable the efficient annotation of compounds in complex systems, laying the groundwork for the preliminary identification of pharmacologically active constituents of Lanbuzheng.

LC-QTOF/MS-Based Profiling of the Phytochemicals in Ice Plant (Mesembryanthemum crystallinum) and Their Bioactivities.

Recent assessments of the correlations between food and medicine underscore the importance of functional foods in disease prevention and management. Functional foods offer health benefits beyond basic nutrition, with fresh fruits and vegetables being particularly prominent because of their rich polyphenol content. In this study, we elucidated the phytochemicals in ice plant (Mesembryanthemum crystallinum), a globally consumed vegetable, using an LC-QTOF/MS-based untargeted detection method. The phytochemicals were clustered based on their structural similarity using molecular networking and annotated using the in silico tool for network annotation propagation. To identify the bioactive compounds, eight compounds were isolated from ice plant extracts. These compounds were identified using extensive spectroscopic methods, including 1H and 13C nuclear magnetic resonance (NMR) spectroscopy. Additionally, we evaluated the antioxidant and anti-inflammatory activities of all the isolates. Among the tested compounds, three showed antioxidant activity and all eight showed anti-inflammatory activity, demonstrating the potential of ice plant as a functional food.

Enhancing Single-Frame Supervision for Better Temporal Action Localization.

Temporal action localization aims to identify the boundaries and categories of actions in videos, such as scoring a goal in a football match. Single-frame supervision has emerged as a labor-efficient way to train action localizers as it requires only one annotated frame per action. However, it often suffers from poor performance due to the lack of precise boundary annotations. To address this issue, we propose a visual analysis method that aligns similar actions and then propagates a few user-provided annotations (e.g., boundaries, category labels) to similar actions via the generated alignments. Our method models the alignment between actions as a heaviest path problem and the annotation propagation as a quadratic optimization problem. As the automatically generated alignments may not accurately match the associated actions and could produce inaccurate localization results, we develop a storyline visualization to explain the localization results of actions and their alignments. This visualization facilitates users in correcting wrong localization results and misalignments. The corrections are then used to improve the localization results of other actions. The effectiveness of our method in improving localization performance is demonstrated through quantitative evaluation and a case study.

Phenylhydrazone Alkaloids from the Deep-Sea Cold Seep Derived Fungus Talaromyces amestolkiae HDN21-0307.

Alkaloids with a phenylhydrazone architecture are rarely found in nature. Four unusual phenylhydrazone alkaloids named talarohydrazones A-D (1-4) were isolated from the deep-sea cold seep derived fungus Talaromyces amestolkiae HDN21-0307 using the one strain-many compounds (OSMAC) approach and MS/MS-based molecular networking (MN) combined with network annotation propagation (NAP) and the unsupervised substructure annotation method MS2LDA. Their structures were elucidated by spectroscopic data analysis, single-crystal X-ray diffraction, and quantum chemical calculations. Talarohydrazone A (1) possessed an unusual skeleton combining 2,4-pyridinedione and phenylhydrazone. Talarohydrazone B (2) represents the first natural phenylhydrazone-bearing azadophilone. Bioactivity evaluation revealed that compound 1 exhibited cytotoxic activity against NCI-H446 cells with an IC50 value of 4.1 μM. In addition, compound 1 displayed weak antibacterial activity toward Staphylococcus aureus with an MIC value of 32 μg/mL.

Computational Methods Reveal a Series of Cyclic and Linear Lichenysins and Surfactins from the Vietnamese Marine Sediment-Derived Streptomyces Strain G222.

The Streptomyces strain G222, isolated from a Vietnamese marine sediment, was confidently identified by 16S rRNA gene sequencing. Its AcOEt crude extract was successfully analyzed using non-targeted LC-MS/MS analysis, and molecular networking, leading to a putative annotation of its chemical diversity thanks to spectral libraries from GNPS and in silico metabolite structure prediction obtained from SIRIUS combined with the bioinformatics tool conCISE (Consensus Annotation Propagation of in silico Elucidations). This dereplication strategy allowed the identification of an interesting cluster of a series of putative cyclic and linear lipopeptides of the lichenysin and surfactin families. Lichenysins (3-7) were isolated from the sub-fraction, which showed significant anti-biofilm activity against Pseudomonas aeruginosa MUC-N1. Their structures were confirmed by detailed 1D and 2D NMR spectroscopy (COSY, HSQC, HMBC, TOCSY, ROESY) recorded in CD3OH, and their absolute configurations were determined using the modified Marfey's method. The isolated lichenysins showed anti-biofilm activity at a minimum concentration of 100 µM. When evaluated for antibacterial activity against a panel of Gram-positive and Gram-negative strains, two isolated lichenysins exhibited selective activity against the MRSA strain without affecting its growth curve and without membranotropic activity. This study highlights the power of the MS/MS spectral similarity strategy using computational methods to obtain a cross-validation of the annotated molecules from the complex metabolic profile of a marine sediment-derived Streptomyces extract. This work provides the first report from a Streptomyces strain of combined cyclic and linear lichenysins and surfactins, known to be characteristic compounds of the genus Bacillus.

Metabolic profiling of Lomatozona artemisiifolia Baker plants grown in vitro and collected from nature using molecular networking and chemometric analysis

Brazilian Cerrado is recognised as a biodiversity hotspot due to the presence of endemic species with great biological potential. Particularly, Lomatozona artemisiifolia, is a rare species found in the Cerrado region in midwestern Brazil. Efforts have been made for its conservation in the Cerrado, such as the use of in vitro micropropagation, demanding a comparative analysis between grown plants and those collected from nature. For this purpose, we performed the chemical study of L. artemisiifolia by LC-ESI-MS/MS and molecular networking analysis in the Global Natural Products Social Molecular Networking (GNPS) with in silico annotation using Network Annotation Propagation (NAP), which led to the observation of labdane diterpenes and flavonoid subclasses as the most representative specialised metabolites of this plant. In addition, molecular networking and chemometric analysis were correlated, allowing the metabolite profile emerging from field growth and micropropagation conditions to be observed.

APEX: an Annotation Propagation Workflow through Multiple Experimental Networks to Improve the Annotation of New Metabolite Classes in Caenorhabditis elegans.

Spectral similarity networks, also known as molecular networks, are crucial in non-targeted metabolomics to aid identification of unknowns aiming to establish a potential structural relation between different metabolite features. However, too extensive differences in compound structures can lead to separate clusters, complicating annotation. To address this challenge, we developed an automated Annotation Propagation through multiple EXperimental Networks (APEX) workflow, which integrates spectral similarity networks with mass difference networks and homologous series. The incorporation of multiple network tools improved annotation quality, as evidenced by high matching rates of the molecular formula derived by SIRIUS. The selection of manual annotations as the Seed Nodes Set (SNS) significantly influenced APEX annotations, with a higher number of seed nodes enhancing the annotation process. We applied APEX to different Caenorhabditis elegans metabolomics data sets as a proof-of-principle for the effective and comprehensive annotation of glycerophospho N-acyl ethanolamides (GPNAEs) and their glyco-variants. Furthermore, we demonstrated the workflow's applicability to two other, well-described metabolite classes in C. elegans, specifically ascarosides and modular glycosides (MOGLs), using an additional publicly available data set. In summary, the APEX workflow presents a powerful approach for metabolite annotation and identification by leveraging multiple experimental networks. By refining the SNS selection and integrating diverse networks, APEX holds promise for comprehensive annotation in metabolomics research, enabling a deeper understanding of the metabolome.

Mass spectrometry-based untargeted metabolomics approaches for comprehensive structural annotation of bioactive metabolites from bushy cashew (Anacardium humile) fruits

Anacardium humile (bushy cashew) is a native Brazilian plant with substantial pharmacological potential and noteworthy commercial significance in the food industry. This study introduces an untargeted metabolomics approach based on mass spectrometry for the comprehensive structural annotation of bioactive metabolites. The fruits were collected from three distinct sites and subjected to LC-HRMS/MS analysis. A total of eighty-eight compounds were putatively annotated across various metabolite classes. This unveiled a metabolic profile characterized by notable concentrations of polyphenols, including flavonoids, tannins, phenolic acids, and quinones, while aliphatic acids and terpenes were found in limited quantities. Noteworthy, no significant disparities in the metabolic content were observed among the collection sites. The three principal metabolites (peonidin 3-O-glucopyranoside, methylcyanidin and methyldelphinidin) underwent assessment for antioxidant activity via molecular docking analysis, subsequently generating structure-based feature pharmacophores. The process of annotation propagation yielded a comprehensive qualitative appraisal of the A. humile metabolome. The outcomes obtained offer potential candidates for further exploration of their nutraceutical attributes.

Molecular networking-guided isolation strategy of a new C-glycosyl flavone rotamer from Stellaria alsine and evaluation of anti-inflammatory and antioxidant activities.

Stellaria alsine has traditionally been used as both a famine relief food and an alternative medicine in East Asia. Modern pharmacological studies have revealed that S. alsine has various biological effects such as anticancer, anti-hepatoma, anti-inflammatory, and antioxidative effects. However, the anti-inflammatory properties of chemical constituents derived from this plant have not been studied well. To identify potential therapeutic candidate for treating inflammatory diseases such as inflammatory bowel disease (IBD). The distribution of chemical compounds was investigated by Global Natural Product Social (GNPS)-based molecular networking (MN) analysis using UPLC-Orbitrap tandem mass spectrometry. The anti-inflammatory and antioxidative effects of S. alsine extracts and fractions were evaluated by measuring interleukin (IL)-8 and reactive oxygen species (ROS) productions. The active EA layer of S. alsine showed the highest percentage of major compounds by feature-based molecular networking. The top candidate structures of EA fraction were rapidly annotated as flavone C- or O-glycosides via an advanced analysis tool, Network Annotation Propagation (NAP). With the GNPS molecular networking-guided isolation strategy, a new C-glycosyl flavone rotamer (1) was isolated. The structures of the major (1a) and minor (1b) rotational isomers were determined by extensive NMR analysis and MS/MS fragmentation. Finally, the anti-inflammatory activity of 1 was predicted by molecular docking simulations with IL-8 protein. These results suggested that the compound 1 is a potential therapeutic candidate for inflammatory bowel disease (IBD).

Rapid discovery of two unprecedented meroterpenoids from Daphne altaica Pall. using molecular networking integrated with MolNetEnhancer and Network Annotation Propagation

Under the guidance of the approach which integrates molecular networking, MolNetEnhancer and Network Annotation Propagation (NAP), daphnaltaicanoids A and B (1 and 2) with unprecedented 9-oxa-tetracyclo[6.6.1.02,6.08,13]pentadecane and tetracyclo[5.3.0.12,5.24,11]tridecane central frameworks were isolated from Daphne altaica Pall., representing two types of unparalleled meroterpenoid cores. Their structures were elucidated by extensive spectroscopic analysis, nuclear magnetic resonance (NMR) calculations, DP4+ analysis and electronic circular dichroism (ECD) calculations. The plausible biosynthetic pathways for 1 and 2 were postulated. Biologically, 2 exerted potent neuroprotective activities which were superior to trolox at 12.5 and 25 µmol/L. Moreover, 1 and 2 exhibited more noticeable acetylcholinesterase inhibitory activities than donepezil. Molecular docking simulations were performed to explore the intermolecular interaction of compounds 1 and 2 with acetylcholinesterase. The bioactivity evaluation results highlight the prospects of 1 and 2 as a novel category of neurological agents.

Molecular networking-assisted isolation of chlorophenolic glycosides from the rhizomes of Curculigo orchioides and their inhibitory effect on α-glucosidase

Molecular networking analysis and in silico tools, such as Network Annotation Propagation (NAP) and MolNetEnhancer, were applied to explore bioactive constituents present in the ethyl acetate-soluble fraction of the rhizomes of Curculigo orchioides. Among the molecular networks, the most abundant cluster was classified as a phenolic glycoside using the ClassyFire module of MolNetEnhancer. Further, the major node in this cluster was accurately predicted as curculigine A using the in silico fragment analysis tool, NAP. Six undescribed chlorophenolic glycosides (1–6) and 11 known phenolic glycosides were isolated, using molecular networking-assisted isolation methods, and their structures were elucidated using 1D, 2D-NMR and HRESIMS. In particular, the structures of the isolated chlorophenolic glycosides, which have non-protonated aromatic rings, were determined using various NMR experiments, such as 1D-selective NOE, ROESY, and LR-HMBC, and acid hydrolysis. All isolated compounds were examined to determine their inhibitory effects on α-glucosidase and compounds 3, 8, 10, 11, 13, 14, and 16 revealed the IC50 values ranging from 19.6 to 35.5 μM. Their structure-activity relationships were also evaluated based on the analysis of their inhibitory effects and performance of molecular docking simulations.

FracAtlas: A Dataset for Fracture Classification, Localization and Segmentation of Musculoskeletal Radiographs

Digital radiography is one of the most common and cost-effective standards for the diagnosis of bone fractures. For such diagnoses expert intervention is required which is time-consuming and demands rigorous training. With the recent growth of computer vision algorithms, there is a surge of interest in computer-aided diagnosis. The development of algorithms demands large datasets with proper annotations. Existing X-Ray datasets are either small or lack proper annotation, which hinders the development of machine-learning algorithms and evaluation of the relative performance of algorithms for classification, localization, and segmentation. We present FracAtlas, a new dataset of X-Ray scans curated from the images collected from 3 major hospitals in Bangladesh. Our dataset includes 4,083 images that have been manually annotated for bone fracture classification, localization, and segmentation with the help of 2 expert radiologists and an orthopedist using the open-source labeling platform, makesense.ai. There are 717 images with 922 instances of fractures. Each of the fracture instances has its own mask and bounding box, whereas the scans also have global labels for classification tasks. We believe the dataset will be a valuable resource for researchers interested in developing and evaluating machine learning algorithms for bone fracture diagnosis.

Structural annotation, semi-quantification and toxicity prediction of pyrrolizidine alkaloids from functional food: In silico and molecular networking strategy.

Many traditional Chinese herbs contain pyrrolizidine alkaloids (PAs), which have been reported to be toxic to livestock and humans. However, the lack of PAs standards makes it difficult to effectively conduct a risk assessment in the varied components of traditional Chinese medicine. It is necessary to propose a suitable strategy to obtain the representative occurrence data of PAs in complex systems. A comprehensive approach for annotating the structures, concentration, and mutagenicity of PAs in three Chinese herbs has been proposed in this article. First, feature-based molecular networking (FBMN) combined with network annotation propagation (NAP) on the Global Natural Products Social Molecular Networking web platform speeds up the process of annotating PAs found in Chinese herbs. Second, a semi-quantitative prediction model based on the quantitative structure and ionization intensity relationship (QSIIR) is used to forecast the amounts of PAs in complex substrates. Finally, the T.E.S.T. was used to provide predictions regarding the mutagenicity of annotated PAs. The goal of this study was to develop a strategy for combining the results of several computer models for PA screening to conduct a comprehensive analysis of PAs, which is a crucial step in risk assessment of unknown PAs in traditional Chinese herbal preparations.

Molecular Networking-Guided Isolation of a Phenolic Constituent from Prunus mume Seed and Its Antioxidant and Anti-Inflammatory Activities.

Prunus mume (Maesil) is used in health foods and alternative medicine in Korea. In the present study, the anti-inflammatory and antioxidant effects of phenolics from P. mume seed extracts were examined. First, the biological activities of various P. mume extracts were evaluated, and the profiles of their chemical compounds were investigated by Global Natural Products Social (GNPS)-molecular networking. Among these extracts, fermented Maesil seed extract (FMSE) showed potent anti-inflammatory and antioxidant activity, and demonstrated the presence of phenolic clusters in GNPS-based studies. Thus, the chemical constituents of this extract were further investigated. Subsequently, the chemical composition of the active CH2Cl2 fraction of FMSE was explored using an advanced GNPS analysis tool, MolNetEnhancer. In addition, the molecular structure of compound 1 from the CH2Cl2 fraction was similarly predicted with Network Annotation Propagation (NAP). Finally, the anti-inflammatory and antioxidant effects of compound 1 were confirmed by lipopolysaccharide (LPS)-induced nitric oxide production and DPPH assay. Western blot analysis revealed that compound 1 downregulated the expression of inducible nitric oxide synthase (iNOS) and cyclooxygenase-2 (COX-2) proteins. The molecular docking simulation additionally confirmed significant interactions of 1 with iNOS and COX-2 proteins. Our findings suggested that an integrated GNPS-based approach could prioritize samples in the early fractionation process and improve the accuracy of target compound prediction.

Improving annotation propagation on molecular networks through random walks: introducing ChemWalker.

Annotation of the mass signals is still the biggest bottleneck for the untargeted mass spectrometry analysis of complex mixtures. Molecular networks are being increasingly adopted by the mass spectrometry community as a tool to annotate large-scale experiments. We have previously shown that the process of propagating annotations from spectral library matches on molecular networks can be automated using Network Annotation Propagation (NAP). One of the limitations of NAP is that the information for the spectral matches is only propagated locally, to the first neighbor of a spectral match. Here, we show that annotation propagation can be expanded to nodes not directly connected to spectral matches using random walks on graphs, introducing the ChemWalker python library. Similarly to NAP, ChemWalker relies on combinatorial in silico fragmentation results, performed by MetFrag, searching biologically relevant databases. Departing from the combination of a spectral network and the structural similarity among candidate structures, we have used MetFusion Scoring function to create a weight function, producing a weighted graph. This graph was subsequently used by the random walk to calculate the probability of 'walking' through a set of candidates, departing from seed nodes (represented by spectral library matches). This approach allowed the information propagation to nodes not directly connected to the spectral library match. Compared with NAP, ChemWalker has a series of improvements, on running time, scalability and maintainability and is available as a standalone python package. ChemWalker is freely available at https://github.com/computational-chemical-biology/ChemWalker. ridasilva@usp.br. Supplementary data are available at Bioinformatics online.

Correction to: Improving annotation propagation on molecular networks through random walks: introducing ChemWalker

Correction to: Improving annotation propagation on molecular networks through random walks: introducing ChemWalker

Context-Driven Detection of Invertebrate Species in Deep-Sea Video

Each year, underwater remotely operated vehicles (ROVs) collect thousands of hours of video of unexplored ocean habitats revealing a plethora of information regarding biodiversity on Earth. However, fully utilizing this information remains a challenge as proper annotations and analysis require trained scientists’ time, which is both limited and costly. To this end, we present a Dataset for Underwater Substrate and Invertebrate Analysis (DUSIA), a benchmark suite and growing large-scale dataset to train, validate, and test methods for temporally localizing four underwater substrates as well as temporally and spatially localizing 59 underwater invertebrate species. DUSIA currently includes over ten hours of footage across 25 videos captured in 1080p at 30 fps by an ROV following pre-planned transects across the ocean floor near the Channel Islands of California. Each video includes annotations indicating the start and end times of substrates across the video in addition to counts of species of interest. Some frames are annotated with precise bounding box locations for invertebrate species of interest, as seen in Fig. 1. To our knowledge, DUSIA is the first dataset of its kind for deep sea exploration, with video from a moving camera, that includes substrate annotations and invertebrate species that are present at significant depths where sunlight does not penetrate. Additionally, we present the novel context-driven object detector (CDD) where we use explicit substrate classification to influence an object detection network to simultaneously predict a substrate and species class influenced by that substrate. We also present a method for improving training on partially annotated bounding box frames. Finally, we offer a baseline method for automating the counting of invertebrate species of interest.

Few‐shot learning for plant disease recognition: A review

AbstractMonitoring plant diseases is essential for farmers to secure crop quantity and quality. Deep learning has recently been applied to plant disease recognition to help farmers take prompt and proper actions to prevent reductions in crop quantity and quality. Generally, deep learning requires a large‐scale dataset with supervised information annotated often by specialists. However, because collecting plant disease images in natural environments is difficult and obtaining proper annotations from specialists is costly, deep learning is infeasible for plant disease recognition tasks. Few‐shot learning (FSL) is an alternative for plant disease recognition using prior knowledge. Although FSL has attracted considerable attention, comprehensive reports on the application of FSL methods for plant disease recognition are required. Here, we introduce FSL with its applications in plant disease recognition. We begin with an overview of computer vision tasks using machine learning and FSL. We provide practical examples of FSL applications. Utilizing these practical examples, we describe different approaches for data augmentation and FSL methods of embedding, multitask learning, transfer learning, and meta‐learning. Further, we summarize how models are optimized for performance with reference to existing studies. Finally, the advantages and disadvantages are discussed, along with potential challenges for FSL applications in plant disease recognition.

A Relook into the Flavonoid Chemical Space of Moringa oleifera Lam. Leaves through a Combination of LC-MS and Molecular Networking.

Moringa oleifera Lam. is a functional tree that is known to produce a variety of metabolites with purported pharmacological activities. It is frequently called the "miracle tree" due to its utilization in numerous nutraceutical and pharmacological contexts. This study was aimed at studying the chemical space of M. oleifera leaf extracts through molecular networking (MN), a tool that identifies metabolites by classifying them based on their MS-based fragmentation pattern similarities and signals. In this case, a special emphasis was placed on the flavonoid composition. The MN unraveled different molecular families such as flavonoids, carboxylic acids and derivatives, lignin glycosides, fatty acyls, and macrolactams that are found within the plant. In silico annotation tools such as network annotation propagation (NAP) and DEREPLICATOR, an unsupervised substructure identification tool (MS2LDA), and MolNet enhancer were also explored to further compliment the classic molecular networking output within the Global Natural Product Social (GNPS) site. In this study, common flavonoids found within Moringa oleifera were further annotated using MS2LDA. Utilizing computational tools allowed for the discovery of a wide range of structurally diverse flavonoid molecules within M. oleifera leaf extracts. The expansion of the flavonoid chemical repertoire in this plant arises from intricate glycosylation modifications, leading to the creation of structural isomers that manifest as isobaric ions during mass spectrometry (MS) analyses.