A New Fingerprint and Graph Hybrid Neural Network for Predicting Molecular Properties.

Machine learning plays a role in accelerating drug discovery, and the design of effective machine learning models is crucial for accurately predicting molecular properties. Characterizing molecules typically involves the use of molecular fingerprints and molecular graphs. These are input into a multilayer perceptron (MLP) and variants of graph neural networks, such as graph attention networks (GATs). Due to the diverse types and large dimension of fingerprints, models may contain many features that are relatively irrelevant or redundant; meanwhile, although the GAT excels in handling heterogeneous graph tasks, it lacks the ability to extract collaborative information from neighboring nodes, which is crucial in scenarios where it cannot capture the joint influence of adjacent groups on atoms. To overcome these challenges, we introduce a hybrid model, combining improved GAT and MLP. In GAT, the recurrent neural network is employed to capture collaborative information. To address the dimensionality issue, we propose a feature selection algorithm, which is based on the principle of maximizing relevance while minimizing redundancy. Through experiments on 13 public data sets and 14 breast cell lines, our model demonstrates superior performance compared to state-of-the-art deep learning and traditional machine learning algorithms. Additionally, a series of ablation experiments were conducted to demonstrate the advantages of our improved version, as well as its antinoise capability and interpretability. These results indicate that our model holds promising prospects for practical applications.

Tumor Cell-Specific Nuclear Targeting of Functionalized Graphene Quantum Dots In Vivo.

Specific targeting of tumor tissues is essential for tumor imaging and therapeutics but remains challenging. Here, we report an unprecedented method using synthetic sulfonic-graphene quantum dots (sulfonic-GQDs) to exactly target the cancer cell nuclei in vivo without any bio- ligand modification, with no intervention in cells of normal tissues. The key factor for such selectivity is the high interstitial fluid pressure (IFP) in tumor tissues, which allows the penetration of sulfonic-GQDs into the plasma membrane of tumor cells. In vitro, the sulfonic-GQDs are repelled out of the cell membrane because of the repulsive force between negatively charged sulfonic-GQDs and the cell membranes which contributes to the low distribution in normal tissues in vivo. However, the plasma membrane-crossing process can be activated by incubating cells in ultrathin film culture medium because of the attachment of sulfonic-GQDs on cell memebranes. Molecular dynamics simulations demonstrated that, once transported across the plasma membrane, the negatively charged functional groups of these GQDs will leave the membrane with a self-cleaning function retaining a small enough size to achieve penetration through the nuclear membrane into the nucleus. Our study showed that IFP is a previously unrecognized mechanism for specific targeting of tumor cell nuclei and suggested that sulfonic-GQDs may be developed into novel tools for tumor-specific imaging and therapeutics.

Manipulating biomolecules with aqueous liquids confined within single-walled nanotubes.

Confinement of molecules inside nanoscale pores has become an important method for exploiting new dynamics not happening in bulk systems and for fabricating novel structures. Molecules that are encapsulated in nanopores are difficult to control with respect to their position and activity. On the basis of molecular dynamics simulations, we have achieved controllable manipulation, both in space and time, of biomolecules with aqueous liquids inside a single-walled nanotube by using an external charge or a group of external charges. The remarkable manipulation abilities are attributed to the single-walled structure of the nanotube that the electrostatic interactions of charges inside and outside the single-walled nanotube are strong enough, and the charge-induced dipole-orientation ordering of water confined in the nanochannel so that water has a strong interaction with the external charge. These designs are expected to serve as lab-in-nanotube for the interactions and chemical reactions of molecules especially biomolecules, and have wide applications in nanotechnology and biotechnology.