Abstract Cytochrome P450 (CYP)–ligand interaction scoring plays a critical role in drug discovery and disease research by distinguishing binders from non-binders based on experimental or simulated CYP–ligand complexes. Accurate scoring of these interactions can help optimize drug dosing, efficacy, and safety. In this study, we develop robust embedding models that generate informative vector representations of CYP–ligand interactions. We focus on evaluating the transferability of these embeddings for downstream predictive tasks, particularly for the underrepresented isoforms such as CYP2B6 and CYP2C8. Using a publicly available dataset covering multiple CYP isoforms, we constructed CYP–ligand complexes via molecular docking and assigned binary activity labels based on pIC50 thresholds. A graph neural network (GNN) architecture was employed to learn structure-based, graph-level embeddings suitable for inhibitor classification. These embeddings were then used as input features for conventional machine learning algorithms, including tree-based ensemble methods and kernel models. Our approach eliminates the need for expert feature engineering and provides generalizable, automated descriptors of CYP–ligand interactions that can support a wide range of predictive tasks in cheminformatics.

BACKGROUND: Predicting of chemical compounds is one of the fundamental tasks in bioinformatics and chemoinformatics, because it contributes to various applications in metabolic engineering and drug discovery. The recent rapid growth of the amount of available data has enabled applications of computational approaches such as statistical modeling and machine learning method. Both a set of chemical interactions and chemical compound structures are represented as graphs, and various graph-based approaches including graph convolutional neural networks have been successfully applied to chemical network prediction. However, there was no efficient method that can consider the two different types of graphs in an end-to-end manner. RESULTS: We give a new formulation of the chemical network prediction problem as a link prediction problem in a graph of graphs (GoG) which can represent the hierarchical structure consisting of compound graphs and an inter-compound graph. We propose a new graph convolutional neural network architecture called dual graph convolutional network that learns compound representations from both the compound graphs and the inter-compound network in an end-to-end manner. CONCLUSIONS: Experiments using four chemical networks with different sparsity levels and degree distributions shows that our dual graph convolution approach achieves high prediction performance in relatively dense networks, while the performance becomes inferior on extremely-sparse networks.

Molecular graphs are one of the established representations for small molecules, and even steric or electronic information can be encoded as node and edge features. Naturally, graph neural networks have been intensively investigated to solve various chemical problems at molecular levels. However, it remains unclear how to encode relevant chemical information into graphs. We investigate this problem by proposing three models of graph neural networks with self-attention mechanisms at different levels to adaptively select relevant chemical information for each input. Using neural graph fingerprint (NFP) as a baseline, we introduce three types of attention mechanisms on the top of NFPs. Our experimental evaluations suggest that introducing these self-attention mechanisms contributes to not only improving the prediction accuracy but also providing quantitative interpretation using obtained attention coefficients.

Triple-negative (TN) breast cancer (BC) patients have limited treatment options and poor prognosis even after extant treatments and standard chemotherapeutic regimens. Linking TN patients to clinically known phenotypes with appropriate treatments is vital. Location-specific sequence variants are expected to be useful for this purpose by identifying subgroups within a disease population. Single gene mutational signatures have been widely reported, with related phenotypes in literature. We thoroughly survey currently available mutations (and mutated genes), linked to BC phenotypes, to demonstrate their limited performance as sole predictors/biomarkers to assign phenotypes to patients. We then explore mutational combinations, as a pilot study, using The Cancer Genome Atlas Research Network mutational data of BC and three machine learning methods: association rules (limitless arity multiple procedure), decision tree and hierarchical disjoint clustering. The study results in a patient classification scheme through combinatorial mutations in Phosphatidylinositol-4,5-Bisphosphate 3-Kinase and tumor protein 53, being consistent with all three methods, implying its validity from a diverse viewpoint. However, it would warrant further research to select multi-gene signatures to identify phenotypes specifically and be clinically used routinely.

Supervised learning over graphs is an intrinsically difficult problem: simultaneous learning of relevant features from the complete subgraph feature set, in which enumerating all subgraph features occurring in given graphs is practically intractable due to combinatorial explosion. We show that 1) existing graph supervised learning studies, such as Adaboost, LPBoost, and LARS/LASSO, can be viewed as variations of a branch-and-bound algorithm with simple bounds, which we call Morishita-Kudo bounds; 2) We present a direct sparse optimization algorithm for generalized problems with arbitrary twice-differentiable loss functions, to which Morishita-Kudo bounds cannot be directly applied; 3) We experimentally showed that i) our direct optimization method improves the convergence rate and stability, and ii) L1-penalized logistic regression (L1-LogReg) by our method identifies a smaller subgraph set, keeping the competitive performance, iii) the learned subgraphs by L1-LogReg are more size-balanced than competing methods, which are biased to small-sized subgraphs.

Computationally predicting drug-target interactions is useful to select possible drug (or target) candidates for further biochemical verification. We focus on machine learning-based approaches, particularly similarity-based methods that use drug and target similarities, which show relationships among drugs and those among targets, respectively. These two similarities represent two emerging concepts, the chemical space and the genomic space. Typically, the methods combine these two types of similarities to generate models for predicting new drug-target interactions. This process is also closely related to a lot of work in pharmacogenomics or chemical biology that attempt to understand the relationships between the chemical and genomic spaces. This background makes the similarity-based approaches attractive and promising. This article reviews the similarity-based machine learning methods for predicting drug-target interactions, which are state-of-the-art and have aroused great interest in bioinformatics. We describe each of these methods briefly, and empirically compare these methods under a uniform experimental setting to explore their advantages and limitations.

The output of frequent pattern mining is a huge number of frequent patterns, which are very redundant, causing a serious problem in understandability. We focus on mining frequent subgraphs for which well-considered approaches to reduce the redundancy are limited because of the complex nature of graphs. Two known, standard solutions are closed and maximal frequent subgraphs, but closed frequent subgraphs are still redundant and maximal frequent subgraphs are too specific. A more promising solution is δ -tolerance closed frequent subgraphs, which decrease monotonically in δ , being equal to maximal frequent subgraphs and closed frequent subgraphs for δ =0 and 1, respectively. However, the current algorithm for mining δ -tolerance closed frequent subgraphs is a naive, two-step approach in which frequent subgraphs are all enumerated and then sifted according to δ -tolerance closedness. We propose an efficient algorithm based on the idea of “reverse-search” by which the completeness of enumeration is guaranteed and for which new pruning conditions are incorporated. We empirically demonstrate that our approach significantly reduced the amount of real computation time of two compared algorithms for mining δ -tolerance closed frequent subgraphs, being pronounced more for practical settings.

MOTIVATION: An observed metabolic response is the result of the coordinated activation and interaction between multiple genetic pathways. However, the complex structure of metabolism has meant that a compete understanding of which pathways are required to produce an observed metabolic response is not fully understood. In this article, we propose an approach that can identify the genetic pathways which dictate the response of metabolic network to specific experimental conditions. RESULTS: Our approach is a combination of probabilistic models for pathway ranking, clustering and classification. First, we use a non-parametric pathway extraction method to identify the most highly correlated paths through the metabolic network. We then extract the defining structure within these top-ranked pathways using both Markov clustering and classification algorithms. Furthermore, we define detailed node and edge annotations, which enable us to track each pathway, not only with respect to its genetic dependencies, but also allow for an analysis of the interacting reactions, compounds and KEGG sub-networks. We show that our approach identifies biologically meaningful pathways within two microarray expression datasets using entire KEGG metabolic networks. AVAILABILITY AND IMPLEMENTATION: An R package containing a full implementation of our proposed method is currently available from http://www.bic.kyoto-u.ac.jp/pathway/timhancock.

MOTIVATION: We address the issue of finding a three-way gene interaction, i.e. two interacting genes in expression under the genotypes of another gene, given a dataset in which expressions and genotypes are measured at once for each individual. This issue can be a general, switching mechanism in expression of two genes, being controlled by categories of another gene, and finding this type of interaction can be a key to elucidating complex biological systems. The most suitable method for this issue is likelihood ratio test using logistic regressions, which we call interaction test, but a serious problem of this test is computational intractability at a genome-wide level. RESULTS: We developed a fast method for this issue which improves the speed of interaction test by around 10 times for any size of datasets, keeping highly interacting genes with an accuracy of approximately 85%. We applied our method to approximately 3 x 10(8) three-way combinations generated from a dataset on human brain samples and detected three-way gene interactions with small P-values. To check the reliability of our results, we first conducted permutations by which we can show that the obtained P-values are significantly smaller than those obtained from permuted null examples. We then used GEO (Gene Expression Omnibus) to generate gene expression datasets with binary classes to confirm the detected three-way interactions by using these datasets and interaction tests. The result showed us some datasets with significantly small P-values, strongly supporting the reliability of the detected three-way interactions. AVAILABILITY: Software is available from http://www.bic.kyoto-u.ac.jp/pathway/kayano/bioinfo_three-way.html CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

A set of convex bodies including samples of a single class only is used for classification. The convex body is defined by some facets (hyper-planes) that separate the class from the other classes. This paper describes an algorithm to find a set of such convex bodies efficiently and examine the performance of a classifier using them. The relationship to the support vector machines is also discussed.

MOTIVATION: Pathway knowledge in public databases enables us to examine how individual metabolites are connected via chemical reactions and what genes are implicated in those processes. For two given (sets of) compounds, the number of possible paths between them in a metabolic network can be intractably large. It would be informative to rank these paths in order to differentiate between them. RESULTS: Focusing on adjacent pairwise coexpression, we developed an algorithm which, for a specified k, efficiently outputs the top k paths based on a probabilistic scoring mechanism, using a given metabolic network and microarray datasets. Our idea of using adjacent pairwise coexpression is supported by recent studies that local coregulation is predominant in metabolism. We first evaluated this idea by examining to what extent highly correlated gene pairs are adjacent and how often they are consecutive in a metabolic network. We then applied our algorithm to two examples of path ranking: the paths from glucose to pyruvate in the entire metabolic network of yeast and the paths from phenylalanine to sinapyl alcohol in monolignols pathways of arabidopsis under several different microarray conditions, to confirm and discuss the performance analysis of our method.

We address the issue of clustering numerical vectors with a network. The problem setting is basically equivalent to constrained clustering by Wagstaff and Cardie and semi-supervised clustering by Basu et al., but our focus is more on the optimal combination of two heterogeneous data sources. An application of this setting is web pages which can be numerically vectorized by their contents, e.g. term frequencies, and which are hyperlinked to each other, showing a network. Another typical application is genes whose behavior can be numerically measured and a gene network can be given from another data source.We first define a new graph clustering measure which we call normalized network modularity, by balancing the cluster size of the original modularity. We then propose a new clustering method which integrates the cost of clustering numerical vectors with the cost of maximizing the normalized network modularity into a spectral relaxation problem. Our learning algorithm is based on spectral clustering which makes our issue an eigenvalue problem and uses k-means for final cluster assignments. A significant advantage of our method is that we can optimize the weight parameter for balancing the two costs from the given data by choosing the minimum total cost. We evaluated the performance of our proposed method using a variety of datasets including synthetic data as well as real-world data from molecular biology. Experimental results showed that our method is effective enough to have good results for clustering by numerical vectors and a network.