Prediction of antibacterial compounds by machine learning approaches

Yang, Xue-Gang; Duan, Chaojun; Wang, Min; Yang, Xue; Chen, Yu Zong

doi:10.1002/jcc.21148

Cited by 37 publications

(28 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One of the easiest methods to understand in ML is k‐nearest neighbours (kNN), an approach used somewhat commonly for QSAR and related problems 37,38. It is based on the assumption that molecules close together in descriptor space will have similar properties 8.…”

Section: Methodsmentioning

confidence: 99%

Greedy and Linear Ensembles of Machine Learning Methods Outperform Single Approaches for QSPR Regression Problems

Kew

Mitchell

2015

Molecular Informatics

View full text Add to dashboard Cite

The application of Machine Learning to cheminformatics is a large and active field of research, but there exist few papers which discuss whether ensembles of different Machine Learning methods can improve upon the performance of their component methodologies. Here we investigated a variety of methods, including kernel-based, tree, linear, neural networks, and both greedy and linear ensemble methods. These were all tested against a standardised methodology for regression with data relevant to the pharmaceutical development process. This investigation focused on QSPR problems within drug-like chemical space. We aimed to investigate which methods perform best, and how the 'wisdom of crowds' principle can be applied to ensemble predictors. It was found that no single method performs best for all problems, but that a dynamic, well-structured ensemble predictor would perform very well across the board, usually providing an improvement in performance over the best single method. Its use of weighting factors allows the greedy ensemble to acquire a bigger contribution from the better performing models, and this helps the greedy ensemble generally to outperform the simpler linear ensemble. Choice of data pre-processing methodology was found to be crucial to performance of each method too.

show abstract

Section: Methodsmentioning

confidence: 99%

Greedy and Linear Ensembles of Machine Learning Methods Outperform Single Approaches for QSPR Regression Problems

Kew

Mitchell

2015

Molecular Informatics

View full text Add to dashboard Cite

show abstract

“…To check the accuracy of the force field for the non-standard Fmoc segment, the structure of Fmoc in each peptide with one water molecule present was fully optimized by applying the B3LYP method and the 6-31G+(d,p) basis set [43][44][45] in the Gaussian 09 program [46]. The optimized geometry was characterized as a true relative energy minimum by a frequency calculation.…”

Section: Dft Studymentioning

confidence: 99%

Molecular dynamics simulation and conformational analysis of some catalytically active peptides

Honarparvar

Skelton

2015

J Mol Model

View full text Add to dashboard Cite

The design of stable and inexpensive artificial enzymes with potent catalytic activity is a growing field in peptide science. The first step in this design process is to understand the key factors that can affect the conformational preference of an enzyme and correlate them with its catalytic activity. In this work, molecular dynamics simulations in explicit water of two catalytically active peptides (peptide 1: Fmoc-Phe1-Phe2-His-CONH2; peptide 2: Fmoc-Phe1-Phe2-Arg-CONH2) were performed at temperatures of 300, 400, and 500 K. Conformational analysis of these peptides using Ramachandran plots identified the secondary structures of the amino acid residues involved (Phe1, Phe2, His, Arg) and confirmed their conformational flexibility in solution. Furthermore, Ramachandran maps revealed the intrinsic preference of the constituent residues of these compounds for a helical conformation. Long-range interaction distances and radius of gyration (R g) values obtained during 20 ns MD simulations confirmed their tendency to form folded conformations. Results showed a decrease in side-chain (Phe1, Phe2, His ring, and Arg) contacts as the temperature was raised from 300 to 400 K and then to 500 K. Finally, the radial distribution functions (RDF) of the water molecules around the nitrogen atoms in the catalytically active His and Arg residues of peptide 1 and peptide 2 revealed that the strongest water-peptide interaction occurred with the arginine nitrogen atoms in peptide 2. Our results highlight differences in the secondary structures of the two peptides that can be explained by the different arrangement of water molecules around the nitrogen atoms of Arg in peptide 2 as compared to the arrangement of water molecules around the nitrogen atoms of His in peptide 1. The results of this work thus provide detailed insight into peptide conformations which can be exploited in the future design of peptide analogs.

show abstract

“…Structural diversity of a set of molecules can be evaluated by D(A), which is the average value of the dissimilarity between all the pairwise molecules in the data set A [28,29].…”

Section: Measurement Of Structural Diversity Of Moleculesmentioning

confidence: 99%

Computer‐aided prediction of toxicity with substructure pattern and random forest

Cao

Yang

Zhao

et al. 2012

Journal of Chemometrics

View full text Add to dashboard Cite

Toxicity of chemicals induced by different factors is an important consideration, especially during the drug research and development process. Thus, there is urgent need to develop computationally effective models that can predict the toxicity or adverse effects of chemicals for a specific class of chemicals. In this study, random forest (RF) was used to classify five toxicity data sets from Distributed Structure-Searchable Toxicity database network, using substructure fingerprints calculated directly from simple molecular structure. Three model validation approaches, out-of-bag validation incorporated in RF, fivefold cross-validation, and an independent validation set, were used for assessing the prediction capability of our models. The chemical space analysis of data sets was explored by multidimensional scaling plots, and outlying molecules were also detected by the proximity measure in RF. At the same time, the important substructure fingerprints, recognized by the RF technique, gave some insights into the structure features related to toxicity of chemicals. The results obtained showed that these in silico classification models with substructure patterns and RF are applicable for potential toxicity prediction of chemical compounds.

show abstract

Prediction of antibacterial compounds by machine learning approaches

Cited by 37 publications

References 55 publications

Greedy and Linear Ensembles of Machine Learning Methods Outperform Single Approaches for QSPR Regression Problems

Greedy and Linear Ensembles of Machine Learning Methods Outperform Single Approaches for QSPR Regression Problems

Molecular dynamics simulation and conformational analysis of some catalytically active peptides

Computer‐aided prediction of toxicity with substructure pattern and random forest

Contact Info

Product

Resources

About