scholarly journals Antimicrobial Activity of Quasi-Enantiomeric Cinchona Alkaloid Derivatives and Prediction Model Developed by Machine Learning

Antibiotics ◽  
2021 ◽  
Vol 10 (6) ◽  
pp. 659
Author(s):  
Alma Ramić ◽  
Mirjana Skočibušić ◽  
Renata Odžak ◽  
Ana Čipak Gašparović ◽  
Lidija Milković ◽  
...  

Bacterial infections that do not respond to current treatments are increasing, thus there is a need for the development of new antibiotics. Series of 20 N-substituted quaternary salts of cinchonidine (CD) and their quasi-enantiomer cinchonine (CN) were prepared and their antimicrobial activity was assessed against a diverse panel of Gram-positive and Gram-negative bacteria. All tested compounds showed good antimicrobial potential (minimum inhibitory concentration (MIC) values 1.56 to 125.00 μg/mL), proved to be nontoxic to different human cell lines, and did not influence the production of reactive oxygen species (ROS). Seven compounds showed very strong bioactivity against some of the tested Gram-negative bacteria (MIC for E. coli and K. pneumoniae 6.25 μg/mL; MIC for P. aeruginosa 1.56 μg/mL). To establish a connection between antimicrobial data and potential energy surfaces (PES) of the compounds, activity/PES models using principal components of the disc diffusion assay and MIC and data towards PES data were built. An extensive machine learning procedure for the generation and cross-validation of multivariate linear regression models with a linear combination of original variables as well as their higher-order polynomial terms was performed. The best possible models with predicted R2(CD derivatives) = 0.9979 and R2(CN derivatives) = 0.9873 were established and presented. This activity/PES model can be used for accurate prediction of activities for new compounds based solely on their potential energy surfaces, which will enable wider screening and guided search for new potential leads. Based on the obtained results, N-quaternary derivatives of Cinchona alkaloids proved to be an excellent scaffold for further optimization of novel antibiotic species.

Author(s):  
Sergei Manzhos ◽  
Eita Sasaki ◽  
Manabu Ihara

Abstract We show that Gaussian process regression (GPR) allows representing multivariate functions with low-dimensional terms via kernel design. When using a kernel built with HDMR (High-dimensional model representation), one obtains a similar type of representation as the previously proposed HDMR-GPR scheme while being faster and simpler to use. We tested the approach on cases where highly accurate machine learning is required from sparse data by fitting potential energy surfaces and kinetic energy densities.


2019 ◽  
Vol 150 (24) ◽  
pp. 244113 ◽  
Author(s):  
Gunnar Schmitz ◽  
Ian Heide Godtliebsen ◽  
Ove Christiansen

2020 ◽  
Vol 1 (1) ◽  
pp. 013001 ◽  
Author(s):  
Oliver T Unke ◽  
Debasish Koner ◽  
Sarbani Patra ◽  
Silvan Käser ◽  
Markus Meuwly

Author(s):  
Evan Komp ◽  
Nida Janulaitis ◽  
Stephanie Valleau

Quantum and classical reaction rate constant calculations come at the cost of exploring potential energy surfaces. Due to the “curse of dimensionality”, their evaluation quickly becomes unfeasible as the system...


2018 ◽  
Vol 15 (1) ◽  
pp. 116-126 ◽  
Author(s):  
Zak E. Hughes ◽  
Joseph C. R. Thacker ◽  
Alex L. Wilson ◽  
Paul L. A. Popelier

Sign in / Sign up

Export Citation Format

Share Document