ECOH: An Enzyme Commission number predictor using mutual information and a support vector machine

Matsuta, Yoshihiko; Ito, Masahiro; Tohsato, Yukako

doi:10.1093/bioinformatics/bts700

Cited by 29 publications

(23 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Matched compounds are supposed to take part in the same metabolic reactions. The Enzyme Commission (EC) number is a numerical classification scheme for enzymes, playing a key role in classifying enzymatic reactions[22]. We expected matched compounds to take part in metabolic reactions with similar EC numbers.…”

Section: Resultsmentioning

confidence: 99%

Atom Identifiers Generated by a Graph Coloring Method Enable Compound Harmonization Across Metabolic Databases

Huang

Mitchell

Moseley

2020

Preprint

View full text Add to dashboard Cite

Metabolic flux analysis requires both a reliable metabolic model and metabolic profiles in 15 characterizing metabolic reprogramming. Advances in analytic methodologies enable production 16of high-quality metabolomics datasets capturing isotopic flux. However, useful metabolic models 17 can be difficult to derive due to the lack of relatively complete atom-resolved metabolic networks 18 for a variety of organisms, including human. Here, we developed a graph coloring method that 19 creates unique identifiers for each atom in a compound facilitating construction of an atom-resolved 20 metabolic network. What is more, this method is guaranteed to generate the same identifier for 21 symmetric atoms, enabling automatic identification of possible additional mappings caused by 22 molecular symmetry. Furthermore, a compound coloring identifier derived from the corresponding 23 atom coloring identifiers can be used for compound harmonization across various metabolic 24 network databases, which is an essential first step in network integration. With the compound 25 coloring identifiers, 8865 correspondences between KEGG and MetaCyc compounds are detected, 26 with 5451 of them confirmed by other identifiers provided by the two databases. In addition, we 27 found that the Enzyme Commission numbers (EC) of reactions can be used to validate possible 28 correspondence pairs, with 1848 unconfirmed pairs validated by commonality in reaction ECs. 29Moreover, we were able to detect various issues and errors with compound representation in KEGG 30 and MetaCyc databases by compound coloring identifiers, demonstrating the usefulness of this 31 methodology for database curation.32 Keywords: Metabolomics; atom-resolved metabolic network; atom identifier; compound identifier; 33 database harmonization; graph theory; common subgraph isomorphism 34 35 1. Introduction 36Metabolic flux analysis is an essential approach to access metabolic phenotypes[1-2] that 37 requires both reliable metabolic profiles as well as metabolic models [3][4][5]. Advances in analytical 38 technologies like mass spectrometry (MS) and nuclear magnetic resonance (NMR) greatly 39 contribute to the detection of thousands of metabolites from biofluids, cells and tissues [6]. 40Application of those analytical techniques to stable isotope resolved metabolomics (SIRM) 41 experiments facilitates production of high-quality metabolomics datasets capturing isotopic flux 42 through cellular and systemic metabolism [7][8]. Now, the challenge is to construct meaningful 43 metabolic models from the corresponding metabolic profiles for downstream metabolic flux 44 2 of 15 analysis. A metabolic network is usually represented by compounds connected via 45 biotransformation routes [9]. Obviously, information at the atom level is not represented in such 46 metabolic networks, making it impractical to derive appropriate metabolic models for SIRM 47 datasets. However, currently there is no relatively complete atom-resolved databases of metabolic 48 networks available for human metabolis...

show abstract

Section: Resultsmentioning

confidence: 99%

Atom Identifiers Generated by a Graph Coloring Method Enable Compound Harmonization Across Metabolic Databases

Huang

Mitchell

Moseley

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…Therefore, the outcome obtained by Jackknife test is always unique for a given dataset. In view of this, many investigators have adopted it to evaluate the accuracies of their classifiers in recent years [25], [26], [27], [28], [29].…”

Section: Methodsmentioning

confidence: 99%

Prediction of Cancer Drugs by Chemical-Chemical Interactions

Lü

Huang

et al. 2014

PLoS ONE

View full text Add to dashboard Cite

Cancer, which is a leading cause of death worldwide, places a big burden on health-care system. In this study, an order-prediction model was built to predict a series of cancer drug indications based on chemical-chemical interactions. According to the confidence scores of their interactions, the order from the most likely cancer to the least one was obtained for each query drug. The 1st order prediction accuracy of the training dataset was 55.93%, evaluated by Jackknife test, while it was 55.56% and 59.09% on a validation test dataset and an independent test dataset, respectively. The proposed method outperformed a popular method based on molecular descriptors. Moreover, it was verified that some drugs were effective to the ‘wrong’ predicted indications, indicating that some ‘wrong’ drug indications were actually correct indications. Encouraged by the promising results, the method may become a useful tool to the prediction of drugs indications.

show abstract

“…This can be problematic because many known enzymatic activities are still missing an associated gene due to annotation errors, the incompleteness of gene sequences (31), and the fact that homology-based methods cannot annotate orphan protein sequences with no or little sequence similarity to known enzymes (3,32). Moreover, sequence similarity methods can provide inaccurate results, as small changes in key residues might greatly alter enzyme functionality (33), and also it is a common observation that vastly different protein sequences can exhibit the same fold and, therefore, have similar catalytic activity even though they look very different (34,35). In addition, these methods are not suitable for the annotation of de novo reactions since current pathway prediction tools only provide information about enzyme catalytic biotransformations and not about their sequences.…”

mentioning

confidence: 99%

“…In addition, these methods are not suitable for the annotation of de novo reactions since current pathway prediction tools only provide information about enzyme catalytic biotransformations and not about their sequences. These shortcomings motivated the development of alternative computational methods based on the structural similarity of reactants and products for identifying candidate protein sequences for orphan enzymatic reactions (30,33,(36)(37)(38)(39)(40). The idea behind these approaches was to assess the similarity of two enzymatic reactions via the similarity of their reaction fingerprints, i.e., the mathematical descriptors of the structural and topological properties of the participating metabolites (41), which could eliminate the problems associated with non-matching or unassigned protein sequences.…”

mentioning

confidence: 99%

“…One class of reaction-fingerprint computational methods compares all of the compounds participating in reactions (40), which includes both reactants and cofactors. The application of this group of methods is restricted to specific enzymatic reactions that do not involve large cofactors (30,33,(36)(37)(38)(39)(40). This is because the structural information of the large cofactors overwhelmingly contributes to the corresponding reconstructed reactionfingerprint, and consequently, reactions with similar cofactors will inaccurately be classified as similar (35)(36)(37)(38).…”

mentioning

confidence: 99%

See 1 more Smart Citation

Assigning enzyme sequences to orphan and novel reactions using knowledge of substrate reactive sites

Hadadi

MohamadiPeyhani

Mišković

et al. 2017

Preprint

View full text Add to dashboard Cite

All rights reserved. No reuse allowed without permission.(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint . http://dx.doi.org/10.1101/210039 doi: bioRxiv preprint first posted online Oct. 27, 2017; 2 AUTHORS SUMMARYThe recent advances in pathway generation tools have resulted in a wealth of de novo hypothetical enzymatic reactions, which lack knowledge of the protein-encoding genes associated with their functionality. Moreover, nearly half of known metabolic enzymes are orphan, i.e., they also lack an associated gene or protein sequence. Proposing genes for catalytic functions of de novo and orphan reactions is critical for their utility in various applications ranging from biotechnology to medicine. In this work, we propose a novel computational method that will bridge the knowledge gap and provide candidate genes for both de novo and orphan reactions. We demonstrate that information about a small chemical structure around the reactive sites of substrates is sufficient to correctly assign genes to the functionality of enzymatic reactions. ABSTRACTThousands of biochemical reactions with characterized biochemical activities are still orphan. Novel reactions predicted by pathway generation tools also lack associated protein sequences and genes.Mapping orphan and novel reactions back to the known biochemistry and proposing genes for their catalytic functions is a daunting problem. We propose a new method, BridgIT, to identify candidate genes and protein sequences for orphan and novel enzymatic reactions. BridgIT introduces, for the first time, the information of the enzyme binding pocket into reaction similarity comparisons. It ascertains the similarity of two reactions by comparing the reactive sites of their substrates and their surrounding structures, along with the structures of the generated products. BridgIT compares orphan and novel reactions to enzymatic reactions with known protein sequences, and then, it proposes protein sequences and genes of the most similar non-orphan reactions as candidates for catalyzing the novel or orphan reactions. We performed BridgIT analysis of orphan reactions from KEGG 2011 (Kyoto Encyclopedia of Genes and Genomes, published in 2011) that became non-orphan in KEGG 2016, and BridgIT correctly predicted enzymes with identical third-and fourth-level EC numbers for 91% and 56% of these reactions, respectively. BridgIT results revealed that it is sufficient to know information about six atoms together with their connecting bonds around the reactive sites of the substrates to match a protein sequence to the catalytic activity of enzymatic reactions with maximal accuracy. Moreover, the same information about only three atoms around the reactive site allowed us to correctly match 87% of the analyzed enzymatic reactions. Finally, we used BridgIT to provide candidate protein sequences for 137000 novel enzymatic reactions from the recently introduced ATLAS of Biochemistry...

show abstract

ECOH: An Enzyme Commission number predictor using mutual information and a support vector machine

Abstract: ECOH is freely available via the Internet at http://www.bioinfo.sk.ritsumei.ac.jp/apps/ecoh/. This program only works on 32-bit Windows.

Cited by 29 publications

References 32 publications

Atom Identifiers Generated by a Graph Coloring Method Enable Compound Harmonization Across Metabolic Databases

Atom Identifiers Generated by a Graph Coloring Method Enable Compound Harmonization Across Metabolic Databases

Prediction of Cancer Drugs by Chemical-Chemical Interactions

Assigning enzyme sequences to orphan and novel reactions using knowledge of substrate reactive sites

Contact Info

Product

Resources

About