Ionic conductive hydrogels with good conductivity and biocompatibility have become one of the research highlights in the field of wearable flexible sensors and supercapacitors. In this work, poly(methacrylic acid–methyl methacrylate)-reinforced poly(sodium acrylate–vinyl phosphonic acid) composite hydrogels (P(AAS-VPA)/PMMS) were designed and tested for strain sensor or supercapacitor applications. The results showed recoverability for 20 cycles of tension and compression experiments, an excellent breaking strain of 2079%, and ionic conductivity of 0.045 S·cm⁻¹, demonstrating strong support for the application of the P(AAS-VPA)/PMMS hydrogel in strain sensors and supercapacitors. The composite hydrogel exhibited outstanding sensing and monitoring capability with high sensitivity (GF = 4.0). The supercapacitor based on the P(AAS-VPA)/PMMS composite hydrogel showed excellent capacitance performance (area capacitance 100.8 mF·cm⁻² and energy density 8.96 μWh·cm⁻²) at ambient temperature and even −30 °C (25.3 mF·cm⁻² and 2.25 μWh·cm⁻²). The hydrogel has stable electrochemical stability (1000 cycles, Coulomb efficiency > 97%) and exhibits electrochemical properties similar to those in the normal state under different deformations. The excellent results demonstrate the great potential of the P(AAS-VPA)/PMMS composite hydrogel in the field of strain sensors and flexible supercapacitors.

DNA N4-methylcytosine(4mC) plays an important role in numerous biological functions and is a mechanism of particular epigenetic importance. Therefore, accurate identification of the 4mC sites in DNA sequences is necessary to understand the functional mechanism. Although some effective calculation tools have been proposed to identifying DNA 4mC sites, it is still challenging to improve identification accuracy and generalization ability. Therefore, there is a great need to build a computational tool to accurately identify the position of DNA 4mC sites. Hence, this study proposed a novel predictor XGB4mcPred, a predictor for the identification of 4mC sites trained using an extreme gradient boosting algorithm (XGBoost) and DNA sequence information. Firstly, we used the One-Hot encoding on adjacent and spaced nucleotides, dinucleotides, and trinucleotides of the original 4mC site sequences as feature vectors. Then, the importance values of the feature vectors pre-trained by the XGBoost algorithm were used as a threshold to filter redundant features, resulting in a significant improvement in the identification accuracy of the constructed XGB4mcPred predictor to identify 4mC sites. The analysis shows that there is a clear preference for nucleotide sequences between 4mC sites and non-4mC site sequences in six datasets from multiple species, and the optimized features can better distinguish 4mC sites from non-4mC sites. The experimental results of cross-validation and independent tests from six different species show that our proposed predictor XGB4mcPred significantly outperformed other state-of-the-art predictors and was improved to varying degrees compared with other state-of-the-art predictors. Additionally, the user-friendly webserver we used to developed the XGB4mcPred predictor was made freely accessible.