Development of an optimum proliferation medium via the graph kernel statistical analysis method for genetically stable in vitro propagation of endemic Thymus cilicicus (Turkey)

Thymus cilicicus is an endemic Eastern Mediterranean element that has aromatic-medicinal properties. Its natural population spreads across gravelly ground and open rocky areas of South and Southwest Anatolia. The current study on in vitro propagation of T. cilicicus focused deeply on environmental applications such as the development of an optimum medium composition for efficient and genetically stable micropropagation and improved preservation procedures for long-time conservation of elite germplasms for further studies. For this purpose, MS and OM media were used individually and in combination with cytokinins, charcoal, AgNO3, Fe-EDDHA, and H3BO3. The raw data were statistically analyzed via the graph kernel method to optimize the nonlinear relationship between all parameters. The optimal proliferation medium for T. cilicicus was OM supplemented with a combination of 10 g L-1 charcoal and 1 mg L-1 KIN and the calculated averages of the best regeneration rate, the best shoot number and the best shoot length were 96.89%, 3 and 1.24 respectively on this medium. The determination of genetic stability of in vitro grown plants on the optimum medium compositions obtained by the graph kernel method was carried out with the use of the ISSR-PCR technique. All the ISSR primers produced a total of 192 reproductive band profiles, none of which were polymorphic. Furthermore, the micropropagated plants were successfully rooted and acclimatized to greenhouse conditions. In this study, we present a graph kernel multiple propagation index which considers all the possible parameters needing to be analyzed. Such an index is used for the first time for the determination of the optimum proliferation medium.

A Comparative Study of Marginalized Graph Kernel and Message Passing Neural Network

<p>This work presents a state-of-the-art hybrid kernel for molecular property predictions. The hybrid kernel consists of a marginalized graph kernel that operates on molecular graphs and radial basis function kernels that operate on global molecular features. Direct message passing neural network (D-MPNN) with global molecular features is used as strong baselines. After using Bayesian optimization to find the optimal hyperparameters, we benchmark the models on 11 publicly available data sets. Our results show that the prediction of the graph kernel is correlated to the prediction of D-MPNN, which indicates that the molecular representation learned from D-MPNN is very close to the reproducing kernel Hilbert space generated by the hybrid kernel. These results may provide clues for research on the interpretability of graph neural networks. In addition, ensembling the graph kernel models with D-MPNN is the best. The advantage of D-MPNN lies in computational efficiency, and the advantage of the graph kernel model lies in the inherent uncertainty qualification of Gaussian process regression. All codes for graph kernel machines used in this work can be found at <a href="https://github.com/Xiangyan93/Chem-Graph-Kernel-Machine">https://github.com/Xiangyan93/Chem-Graph-Kernel-Machine</a>.</p>

A Comparative Study of Marginalized Graph Kernel and Message Passing Neural Network

<p>This work presents a state-of-the-art hybrid kernel for molecular property predictions. The hybrid kernel consists of a marginalized graph kernel that operates on molecular graphs and radial basis function kernels that operate on global molecular features. Direct message passing neural network (D-MPNN) with global molecular features is used as strong baselines. After using Bayesian optimization to find the optimal hyperparameters, we benchmark the models on 11 publicly available data sets. Our results show that the prediction of the graph kernel is correlated to the prediction of D-MPNN, which indicates that the molecular representation learned from D-MPNN is very close to the reproducing kernel Hilbert space generated by the hybrid kernel. These results may provide clues for research on the interpretability of graph neural networks. In addition, ensembling the graph kernel models with D-MPNN is the best. The advantage of D-MPNN lies in computational efficiency, and the advantage of the graph kernel model lies in the inherent uncertainty qualification of Gaussian process regression. All codes for graph kernel machines used in this work can be found at <a href="https://github.com/Xiangyan93/Chem-Graph-Kernel-Machine">https://github.com/Xiangyan93/Chem-Graph-Kernel-Machine</a>.</p>