Effectiveness of Artificial Intelligence (AI) in Improving Pupils' Deep Learning in Primary School Mathematics Teaching in Fujian Province.

Several primary school students in Fujian Province have perceived studying mathematics as challenging. To deal with this issue, computer technology advancements, specifically artificial intelligence (AI), present an opportunity to evaluate individual students' learning challenges and give individualized support to optimize their success in mathematics classes. It is also possible to use virtual reality (VR) to assist learners in acquiring complex mathematical and logical ideas and to lessen learners' mistakes. As a result, researchers, particularly beginners, are missing out on a complete perspective of the study of AI in teaching mathematics. That is why we are exploring the role of AI in math education by developing a "fuzzy-based tweakable convolution neural network with a long short-term memory (FT-CNN-LSTM-AM)" method. For this investigation, the students' datasets are taken and educated by mathematical teaching via the application of AI. The proposed method is utilized to predict the students' performance in mathematical education. A grey wolf optimizer is employed to boost the effectiveness of the proposed method. Furthermore, the performance of the proposed method is analyzed and compared with existing approaches to gain the highest reliability.

A Novel Scalarized Scaffold Hopping Algorithm with Graph-Based Variational Autoencoder for Discovery of JAK1 Inhibitors.

We have developed a graph-based Variational Autoencoder with Gaussian Mixture hidden space (GraphGMVAE), a deep learning approach for controllable magnitude of scaffold hopping in generative chemistry. It can effectively and accurately generate molecules from a given reference compound, with excellent scaffold novelty against known molecules in the literature or patents (97.9% are novel scaffolds). Moreover, a pipeline for prioritizing the generated compounds was also proposed to narrow down our validation focus. In this work, GraphGMVAE was validated by rapidly hopping the scaffold from FDA-approved upadacitinib, which is an inhibitor of human Janus kinase 1 (JAK1), to generate more potent molecules with novel chemical scaffolds. Seven compounds were synthesized and tested to be active in biochemical assays. The most potent molecule has 5.0 nM activity against JAK1 kinase, which shows that the GraphGMVAE model can design molecules like how a human expert does but with high efficiency and accuracy.

Discovery of promising FtsZ inhibitors by E-pharmacophore, 3D-QSAR, molecular docking study, and molecular dynamics simulation.

Asbtract Filamentous temperature-sensitive protein Z (FtsZ), playing a key role in bacterial cell division, is regarded as a promising target for the design of antimicrobial agent. This study is looking for potential high-efficiency FtsZ inhibitors. Ligand-based pharmacophore and E-pharmacophore, virtual screening and molecular docking were used to detect promising FtsZ inhibitors, and molecular dynamics simulation was used to study the stability of protein-ligand complexes in this paper. Sixty-three inhibitors from published literatures with pIC50 ranging from 2.483 to 5.678 were collected to develop ligand-based pharmacophore model. 4DXD bound with 9PC was selected to develop the E-pharmacophore model. The pharmacophore models validated by test set method and decoy set were employed for virtual screening to exclude inactive compounds against ZINC database. After molecular docking, ADME analysis, IFD docking and MM-GBSA, 8 hits were identified as potent FtsZ inhibitors. A 50 ns molecular dynamics simulation was implemented on the compounds to assess the stability between potent inhibitors and FtsZ. The results indicated that the candidate compounds had a high docking score and were strongly combined with FtsZ by forming hydrogen bonding interactions with key amino acid residues, and van der Waals forces and hydrophobic interactions had significant contribution to the stability of the binding. Molecular dynamics simulation results showed that the protein-ligand compounds performed well in both the stability and flexibility of the simulation process.

Pharmacophore modeling, multiple docking, and molecular dynamics studies on Wee1 kinase inhibitors.

Wee1-like protein kinase (Wee1) is a tyrosine kinase that regulates the G2 checkpoint and prevents entry into mitosis in response to DNA damage. Based on a series of signaling pathways initiated by Wee1, Wee1 has been recognized as a potential target for cancer therapy. To discover potent Wee1 inhibitors with novel scaffolds, ligand-based pharmacophore model has been built based on 101 known Wee1 inhibitors. Then the best pharmacophore model, AADRRR.340, with good partial least square (PLS) statistics (R2 = 0.9212, Q2 = 0.7457), was selected and validated. The validated model was used as a three-dimensional (3D) search query for databases virtual screening. The filtered molecules were further analyzed and refined by Lipinski's rule of 5, multiple docking procedures (high throughput virtual screening (HTVS), standard precision (SP), genetic optimization for ligand docking (GOLD), extra precision (XP), and unique quantum polarized ligand docking (QPLD)); absorption, distribution, metabolism, excretion, and toxicity (ADMET) screening; and the Prime/molecular mechanics generalized born surface area (MM-GBSA) method binding free energy calculations. Eight leads were identified as potential Wee1 inhibitors, and a 50 ns molecular dynamics (MD) simulation was carried out for top four inhibitors to predict the stability of ligand-protein complex. Molecular mechanics Poisson-Boltzmann surface area (MM-PBSA) based on MD simulation and the energy contribution per residue to the binding energy were calculated. In the end, three hits with good stabilization and affinity to protein were identified. Communicated by Ramaswamy H. Sarma.

Discover potential inhibitors for PFKFB3 using 3D-QSAR, virtual screening, molecular docking and molecular dynamics simulation.

The 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase-3 (PFKFB3) is a master regulator of glycolysis in cancer cells by synthesizing fructose-2,6-bisphosphate (F-2,6-BP), a potent allosteric activator of phosphofructokinase-1 (PFK-1), which is a rate-limiting enzyme of glycolysis. PFKFB3 is an attractive target for cancer treatment. It is valuable to discover promising inhibitors by using 3D-QSAR pharmacophore modeling, virtual screening, molecular docking and molecular dynamics simulation. Twenty molecules with known activity were used to build 3D-QSAR pharmacophore models. The best pharmacophore model was ADHR called Hypo1, which had the highest correlation value of 0.98 and the lowest RMSD of 0.82. Then, the Hypo1 was validated by cost value method, test set method and decoy set validation method. Next, the Hypo1 combined with Lipinski's rule of five and ADMET properties were employed to screen databases including Asinex and Specs, total of 1,048,159 molecules. The hits retrieved from screening were docked into protein by different procedures including HTVS, SP and XP. Finally, nine molecules were picked out as potential PFKFB3 inhibitors. The stability of PFKFB3-lead complexes was verified by 40 ns molecular dynamics simulation. The binding free energy and the energy contribution of per residue to the binding energy were calculated by MM-PBSA based on molecular dynamics simulation.