Structure-aware deep model for MHC-II peptide binding affinity prediction.

The prediction of major histocompatibility complex (MHC)-peptide binding affinity is an important branch in immune bioinformatics, especially helpful in accelerating the design of disease vaccines and immunity therapy. Although deep learning-based solutions have yielded promising results on MHC-II molecules in recent years, these methods ignored structure knowledge from each peptide when employing the deep neural network models. Each peptide sequence has its specific combination order, so it is worth considering adding the structural information of the peptide sequence to the deep model training. In this work, we use positional encoding to represent the structural information of peptide sequences and validly combine the positional encoding with existing models by different strategies. Experiments on three datasets show that the introduction of position-coding information can further improve the performance built upon the existing model. The idea of introducing positional encoding to this field can provide important reference significance for the optimization of the deep network structure in the future.

A reinforcement learning algorithm acquires demonstration from the training agent by dividing the task space.

Although reinforcement learning (RL) has made numerous breakthroughs in recent years, addressing reward-sparse environments remains challenging and requires further exploration. Many studies improve the performance of the agents by introducing the state-action pairs experienced by an expert. However, such kinds of strategies almost depend on the quality of the demonstration by the expert, which is rarely optimal in a real-world environment, and struggle with learning from sub-optimal demonstrations. In this paper, a self-imitation learning algorithm based on the task space division is proposed to realize an efficient high-quality demonstration acquire while the training process. To determine the quality of the trajectory, some well-designed criteria are defined in the task space for finding a better demonstration. The results show that the proposed algorithm will improve the success rate of robot control and achieve a high mean Q value per step. The algorithm framework proposed in this paper has illustrated a great potential to learn from a demonstration generated by using self-policy in sparse environments and can be used in reward-sparse environments where the task space can be divided.

Upconversion optogenetics-driven biohybrid sensor for infrared sensing and imaging.

Living organisms are far superior to state-of-the-art devices in visual perception as they have evolved a wide number of capabilities that encompass our most advanced technologies. By leveraging the performance of living organisms and directly interfacing them with artificial components, it can use the intricacy and metabolic efficiency of biological visual sensing within artificial machines. Inspired by the molecular basis (transient receptor potential, TRP) for infrared detection of pit-bearing organisms, we propose a TRP-like biohybrid sensor by integrating upconversion nanoparticles (UCNP) and optogenetically engineered cells on a graphene transistor for infrared sensing and imaging. The UCNP converts infrared light irradiation into blue light, the blue light activates the cells expressed with channelrhodopsin-2 (ChR2) and induces transmembrane photocurrent, and the photocurrent is detected by a biocompatible graphene transistor. Stepwise and overall experimental results show that, upon infrared light irradiation, the UCNP can rapidly mediate cellular photocurrents, which further translates into the extra output current of the graphene transistor. More notably, the response speed of the biohybrid sensor is 1∼3 orders of magnitude faster than those of TRPs heterologously expressed in cell lines in the literature, which confirms the response time advantage of the combination of UCNP and ChR2 within the sensor in place of TRPs. The biohybrid sensor can successfully image infrared targets, proving the feasibility of developing bionic infrared sensing devices by biohybrid integration of nonliving nanomaterials and biological components. This work opens up an avenue for biohybrid sensors to develop the bionic infrared vision that promisingly reproduces the functional superiority of natural organisms. STATEMENT OF SIGNIFICANCE: Infrared sensing and imaging have a wide range of military and civilian applications. Organisms have evolved excellent infrared vision with the molecular basis, transient receptor potential (TRP), and the performance is superior to existing state-of-the-art infrared devices. Inspired by this, a TRP-like biohybrid sensor based on upconversion optogenetics and a 2D material-based device is developed for infrared sensing and imaging. The biohybrid sensor has a relatively fast response speed that is 1∼3 orders of magnitude faster than that of the heterologously expressed TRPs, which enables its capability of infrared imaging with a single pixel-based method. This work broadens the spectrum of biohybrid sensing based on engineered cells to infrared, advancing the process of reproducing the excellent infrared detection of organisms.

Unidirectional diphenylalanine nanotubes for dynamically guiding neurite outgrowth.

Neural networks have been culturedin vitroto investigate brain functions and diseases, clinical treatments for brain damage, and device development. However, it remains challenging to form complex neural network structures with desired orientations and connectionsin vitro. Here, we introduce a dynamic strategy by using diphenylalanine (FF) nanotubes for controlling physical patterns on a substrate to regulate neurite-growth orientation in cultivating neural networks. Parallel FF nanotube patterns guide neurons to develop neurites through the unidirectional FF nanotubes while restricting their polarization direction. Subsequently, the FF nanotubes disassemble and the restriction of neurites disappear, and secondary neurite development of the neural network occurs in other direction. Experiments were conducted that use the hippocampal neurons, and the results demonstrated that the cultured neural networks by using the proposed dynamic approach can form a significant cross-connected structure with substantially more lateral neural connections than static substrates. The proposed dynamic approach for neurite outgrowing enables the construction of oriented innervation and cross-connected neural networksin vitroand may explore the way for the bio-fabrication of highly complex structures in tissue engineering.