Hierarchical graph transformer with contrastive learning for protein function prediction.

MOTIVATION: In recent years, high-throughput sequencing technologies have made large-scale protein sequences accessible. However, their functional annotations usually rely on low-throughput and pricey experimental studies. Computational prediction models offer a promising alternative to accelerate this process. Graph neural networks have shown significant progress in protein research, but capturing long-distance structural correlations and identifying key residues in protein graphs remains challenging. RESULTS: In the present study, we propose a novel deep learning model named Hierarchical graph transformEr with contrAstive Learning (HEAL) for protein function prediction. The core feature of HEAL is its ability to capture structural semantics using a hierarchical graph Transformer, which introduces a range of super-nodes mimicking functional motifs to interact with nodes in the protein graph. These semantic-aware super-node embeddings are then aggregated with varying emphasis to produce a graph representation. To optimize the network, we utilized graph contrastive learning as a regularization technique to maximize the similarity between different views of the graph representation. Evaluation of the PDBch test set shows that HEAL-PDB, trained on fewer data, achieves comparable performance to the recent state-of-the-art methods, such as DeepFRI. Moreover, HEAL, with the added benefit of unresolved protein structures predicted by AlphaFold2, outperforms DeepFRI by a significant margin on Fmax, AUPR, and Smin metrics on PDBch test set. Additionally, when there are no experimentally resolved structures available for the proteins of interest, HEAL can still achieve better performance on AFch test set than DeepFRI and DeepGOPlus by taking advantage of AlphaFold2 predicted structures. Finally, HEAL is capable of finding functional sites through class activation mapping. AVAILABILITY AND IMPLEMENTATION: Implementations of our HEAL can be found at https://github.com/ZhonghuiGu/HEAL.

QuoteTarget: A sequence-based transformer protein language model to identify potentially druggable protein targets.

The development of efficient computational methods for drug target protein identification can compensate for the high cost of experiments and is therefore of great significance for drug development. However, existing structure-based drug target protein-identification algorithms are limited by the insufficient number of proteins with experimentally resolved structures. Moreover, sequence-based algorithms cannot effectively extract information from protein sequences and thus display insufficient accuracy. Here, we combined the sequence-based self-supervised pretraining protein language model ESM1b with a graph convolutional neural network classifier to develop an improved, sequence-based drug target protein identification method. This complete model, named QuoteTarget, efficiently encodes proteins based on sequence information alone and achieves an accuracy of 95% with the nonredundant drug target and nondrug target datasets constructed for this study. When applied to all proteins from Homo sapiens, QuoteTarget identified 1213 potential undeveloped drug target proteins. We further inferred residue-binding weights from the well-trained network using the gradient-weighted class activation mapping (Grad-Cam) algorithm. Notably, we found that without any binding site information input, significant residues inferred by the model closely match the experimentally confirmed drug molecule-binding sites. Thus, our work provides a highly effective sequence-based identifier for drug target proteins, as well to yield new insights into recognizing drug molecule-binding sites. The entire model is available at https://github.com/Chenjxjx/drug-target-prediction.

In silico protein function prediction: the rise of machine learning-based approaches.

Proteins function as integral actors in essential life processes, rendering the realm of protein research a fundamental domain that possesses the potential to propel advancements in pharmaceuticals and disease investigation. Within the context of protein research, an imperious demand arises to uncover protein functionalities and untangle intricate mechanistic underpinnings. Due to the exorbitant costs and limited throughput inherent in experimental investigations, computational models offer a promising alternative to accelerate protein function annotation. In recent years, protein pre-training models have exhibited noteworthy advancement across multiple prediction tasks. This advancement highlights a notable prospect for effectively tackling the intricate downstream task associated with protein function prediction. In this review, we elucidate the historical evolution and research paradigms of computational methods for predicting protein function. Subsequently, we summarize the progress in protein and molecule representation as well as feature extraction techniques. Furthermore, we assess the performance of machine learning-based algorithms across various objectives in protein function prediction, thereby offering a comprehensive perspective on the progress within this field.

Accurate mechanical-optical theoretical model of cross-axis sensitivity of an interferometric micro-optomechanical accelerometer.

An interferometric micro-optomechanical accelerometer usually has ultrahigh sensitivity and accuracy. However, cross-axis interference inevitably degrades the performance, including its detection accuracy and output signal contrast. To accurately clarify the influence of cross-axis interference, a modified mechanical-optical theoretical model is established. The rotation of the proof mass and the detected light intensity are quantitatively investigated with a load of cross-axis acceleration. A simulation and experiment are performed to verify the correctness of the theoretical model when the cross-axis acceleration is from 0 to 0.175 g. The results demonstrate that this model has a more than fivefold accuracy increase compared with conventional theoretical models when the cross-axis acceleration is from 0.06 to 0.175 g. In addition, we provide a suppression method to diminish the rotation of the proof mass based on squeeze film air damping, which significantly suppresses the contrast reduction caused by cross-axis interference.

Extending the Validity of Squeeze Film Damping Models with Lower Aspect Ratios.

Squeeze film air damping is a significant factor in the design of MEMS devices owing to its great impact on the dynamic performance of vibrating structures. However, the traditional theoretical results of squeeze film air damping are derived from the Reynolds equation, wherein there exists a deviation from the true results, especially in low aspect ratios. While expensive efforts have been undertaken to prove that this deviation is caused by the neglect of pressure change across the film, a quantitative study has remained elusive. This paper focuses on the investigation of the finite size effect of squeeze film air damping and conducts numerical research using a set of simulations. A modified expression is extended to lower aspect ratio conditions from the original model of squeeze film air damping. The new quick-calculating formulas based on the simulation results reproduce the squeeze film air damping with a finite size effect accurately with a maximum error of less than 1% in the model without a border effect and 10.185% in the compact model with a border effect. The high consistency between the new formulas and simulation results shows that the finite size effect was adequately considered, which offers a previously unattainable precise damping design guide for MEMS devices.

A Temperature Control Method for Microaccelerometer Chips Based on Genetic Algorithm and Fuzzy PID Control.

External temperature changes can detrimentally affect the properties of a microaccelerometer, especially for high-precision accelerometers. Temperature control is the fundamental method to reduce the thermal effect on microaccelerometer chips, although high-performance control has remained elusive using the conventional proportional-integral-derivative (PID) control method. This paper proposes a modified approach based on a genetic algorithm and fuzzy PID, which yields a profound improvement compared with the typical PID method. A sandwiched microaccelerometer chip with a measurement resistor and a heating resistor on the substrate serves as the hardware object, and the transfer function is identified by a self-built measurement system. The initial parameters of the modified PID are obtained through the genetic algorithm, whereas a fuzzy strategy is implemented to enable real-time adjustment. According to the simulation results, the proposed temperature control method has the advantages of a fast response, short settling time, small overshoot, small steady-state error, and strong robustness. It outperforms the normal PID method and previously reported counterparts. This design method as well as the approach can be of practical use and applied to chip-level package structures.

Investigation of a complete squeeze-film damping model for MEMS devices.

Dynamic performance has long been critical for micro-electro-mechanical system (MEMS) devices and is significantly affected by damping. Different structural vibration conditions lead to different damping effects, including border and amplitude effects, which represent the effect of gas flowing around a complicated boundary of a moving plate and the effect of a large vibration amplitude, respectively. Conventional models still lack a complete understanding of damping and cannot offer a reasonably good estimate of the damping coefficient for a case with both effects. Expensive efforts have been undertaken to consider these two effects, yet a complete model has remained elusive. This paper investigates the dynamic performance of vibrated structures via theoretical and numerical methods simultaneously, establishing a complete model in consideration of both effects in which the analytical expression is given, and demonstrates a deviation of at least threefold lower than current studies by simulation and experimental results. This complete model is proven to successfully characterize the squeeze-film damping and dynamic performance of oscillators under comprehensive conditions. Moreover, a series of simulation models with different dimensions and vibration statuses are introduced to obtain a quick-calculating factor of the damping coefficient, thus offering a previously unattainable damping design guide for MEMS devices.