Classification of alkaloids according to the starting substances of their biosynthetic pathways using graph convolutional neural networks.

BACKGROUND: Alkaloids, a class of organic compounds that contain nitrogen bases, are mainly synthesized as secondary metabolites in plants and fungi, and they have a wide range of bioactivities. Although there are thousands of compounds in this class, few of their biosynthesis pathways are fully identified. In this study, we constructed a model to predict their precursors based on a novel kind of neural network called the molecular graph convolutional neural network. Molecular similarity is a crucial metric in the analysis of qualitative structure-activity relationships. However, it is sometimes difficult for current fingerprint representations to emphasize specific features for the target problems efficiently. It is advantageous to allow the model to select the appropriate features according to data-driven decisions for extracting more useful information, which influences a classification or regression problem substantially. RESULTS: In this study, we applied a neural network architecture for undirected graph representation of molecules. By encoding a molecule as an abstract graph and applying "convolution" on the graph and training the weight of the neural network framework, the neural network can optimize feature selection for the training problem. By incorporating the effects from adjacent atoms recursively, graph convolutional neural networks can extract the features of latent atoms that represent chemical features of a molecule efficiently. In order to investigate alkaloid biosynthesis, we trained the network to distinguish the precursors of 566 alkaloids, which are almost all of the alkaloids whose biosynthesis pathways are known, and showed that the model could predict starting substances with an averaged accuracy of 97.5%. CONCLUSION: We have showed that our model can predict more accurately compared to the random forest and general neural network when the variables and fingerprints are not selected, while the performance is comparable when we carefully select 507 variables from 18000 dimensions of descriptors. The prediction of pathways contributes to understanding of alkaloid synthesis mechanisms and the application of graph based neural network models to similar problems in bioinformatics would therefore be beneficial. We applied our model to evaluate the precursors of biosynthesis of 12000 alkaloids found in various organisms and found power-low-like distribution.

KNApSAcK Metabolite Activity Database for retrieving the relationships between metabolites and biological activities.

Databases (DBs) are required by various omics fields because the volume of molecular biology data is increasing rapidly. In this study, we provide instructions for users and describe the current status of our metabolite activity DB. To facilitate a comprehensive understanding of the interactions between the metabolites of organisms and the chemical-level contribution of metabolites to human health, we constructed a metabolite activity DB known as the KNApSAcK Metabolite Activity DB. It comprises 9,584 triplet relationships (metabolite-biological activity-target species), including 2,356 metabolites, 140 activity categories, 2,963 specific descriptions of biological activities and 778 target species. Approximately 46% of the activities described in the DB are related to chemical ecology, most of which are attributed to antimicrobial agents and plant growth regulators. The majority of the metabolites with antimicrobial activities are flavonoids and phenylpropanoids. The metabolites with plant growth regulatory effects include plant hormones. Over half of the DB contents are related to human health care and medicine. The five largest groups are toxins, anticancer agents, nervous system agents, cardiovascular agents and non-therapeutic agents, such as flavors and fragrances. The KNApSAcK Metabolite Activity DB is integrated within the KNApSAcK Family DBs to facilitate further systematized research in various omics fields, especially metabolomics, nutrigenomics and foodomics. The KNApSAcK Metabolite Activity DB could also be utilized for developing novel drugs and materials, as well as for identifying viable drug resources and other useful compounds.

Systematization of the protein sequence diversity in enzymes related to secondary metabolic pathways in plants, in the context of big data biology inspired by the KNApSAcK motorcycle database.

Biology is increasingly becoming a data-intensive science with the recent progress of the omics fields, e.g. genomics, transcriptomics, proteomics and metabolomics. The species-metabolite relationship database, KNApSAcK Core, has been widely utilized and cited in metabolomics research, and chronological analysis of that research work has helped to reveal recent trends in metabolomics research. To meet the needs of these trends, the KNApSAcK database has been extended by incorporating a secondary metabolic pathway database called Motorcycle DB. We examined the enzyme sequence diversity related to secondary metabolism by means of batch-learning self-organizing maps (BL-SOMs). Initially, we constructed a map by using a big data matrix consisting of the frequencies of all possible dipeptides in the protein sequence segments of plants and bacteria. The enzyme sequence diversity of the secondary metabolic pathways was examined by identifying clusters of segments associated with certain enzyme groups in the resulting map. The extent of diversity of 15 secondary metabolic enzyme groups is discussed. Data-intensive approaches such as BL-SOM applied to big data matrices are needed for systematizing protein sequences. Handling big data has become an inevitable part of biology.

KNApSAcK-3D: a three-dimensional structure database of plant metabolites.

Studies on plant metabolites have attracted significant attention in recent years. Over the past 8 years, we have constructed a unique metabolite database, called KNApSAcK, that contains information on the relationships between metabolites and their expressing organism(s). In the present paper, we introduce KNApSAcK-3D, which contains the three-dimensional (3D) structures of all of the metabolic compounds included in the original KNApSAcK database. The 3D structure for each compound was optimized using the Merck Molecular Force Field (MMFF94), and a multiobjective genetic algorithm was used to search extensively for possible conformations and locate the global minimum. The resulting set of structures may be used for docking studies to identify new and potentially unexpected binding sites for target proteins. The 3D structures may also be utilized for more qualitative studies, such as the estimation of biological activities using 3D-QSAR. The database can be accessed via a link from the KNApSAcK Family website (http://kanaya.naist.jp/KNApSAcK_Family/) or directory at http://kanaya.naist.jp/knapsack3d/.

KNApSAcK family databases: integrated metabolite-plant species databases for multifaceted plant research.

A database (DB) describing the relationships between species and their metabolites would be useful for metabolomics research, because it targets systematic analysis of enormous numbers of organic compounds with known or unknown structures in metabolomics. We constructed an extensive species-metabolite DB for plants, the KNApSAcK Core DB, which contains 101,500 species-metabolite relationships encompassing 20,741 species and 50,048 metabolites. We also developed a search engine within the KNApSAcK Core DB for use in metabolomics research, making it possible to search for metabolites based on an accurate mass, molecular formula, metabolite name or mass spectra in several ionization modes. We also have developed databases for retrieving metabolites related to plants used for a range of purposes. In our multifaceted plant usage DB, medicinal/edible plants are related to the geographic zones (GZs) where the plants are used, their biological activities, and formulae of Japanese and Indonesian traditional medicines (Kampo and Jamu, respectively). These data are connected to the species-metabolites relationship DB within the KNApSAcK Core DB, keyed via the species names. All databases can be accessed via the website http://kanaya.naist.jp/KNApSAcK_Family/. KNApSAcK WorldMap DB comprises 41,548 GZ-plant pair entries, including 222 GZs and 15,240 medicinal/edible plants. The KAMPO DB consists of 336 formulae encompassing 278 medicinal plants; the JAMU DB consists of 5,310 formulae encompassing 550 medicinal plants. The Biological Activity DB consists of 2,418 biological activities and 33,706 pairwise relationships between medicinal plants and their biological activities. Current statistics of the binary relationships between individual databases were characterized by the degree distribution analysis, leading to a prediction of at least 1,060,000 metabolites within all plants. In the future, the study of metabolomics will need to take this huge number of metabolites into consideration.