The accurate identification of viral sequences in sequencing data is critical for pathogen surveillance, especially in metagenomic environments where viral reads are ultra-short, noisy, and deeply embedded in host material. Existing methods struggle with generalizability and robustness, particularly on short fragments that are common in real-world sequencing data. In this work, we present Ev02- Virus, a lightweight and scalable framework that combines autoregressive pretraining with bidirectional Transformer fine-tuning to classify viral fragments as short as 105 bp. We curate a large-scale pretraining corpus from the Virus-Host DB and a balanced classification dataset constructed from human RNA-seq samples. Evo2-Virus is trained end-to-end and requires no hand-crafted features or sequence alignment. Extensive experiments show that Evo2-Virus significantly outperforms state-of-the-art models including Seeker, DeepVirFinder, RNA-FM, and RNN-VirSeeker, achieving an Fl-score of 0.889 and AUC of 0.957. The model demonstrates remarkable robustness under synthetic sequence noise and exhibits clear separability between viral and non-viral embed dings in low-dimensional space. Our results highlight Ev02- Virus as a practical and effective solution for ultra-short viral sequence detection, with direct implications for real-world pathogen screening, biosurveillance, and genomic diagnostics.

Histone modifications play a crucial role in transcriptional regulation and are essential targets for genome annotation and gene expression modeling. However, experimentally profiling histone modifications across species, tissues, and cellular states is costly and often impractical. While numerous deep learning models have been proposed to predict histone modifications, most operate at relatively low resolution (128 bp), limiting their utility in fine-scale genomic analysis. In this study, we present MambaHM (Mamba for predicting Histone Modifications), a novel deep learning framework built upon Mamba architecture. MambaHM integrates DNA sequence features with chromatin accessibility data (ATAC-seq) to predict ten histone modifications. Leveraging the linear computational complexity of Mamba's state-space modeling, our model avoids excessive compression of feature length during embedding. This enables the model to retain sufficient contextual information while simultaneously achieving 16-bp resolution, thereby enhancing the granularity and accuracy of prediction. Experiments demonstrate that MambaHM achieves a mean Pearson correlation of 0.836 (± 0.063) on the K562 cell line test set. Furthermore, the model generalizes well across cell types, tissues, and species, with a mean cross-context correlation of$0.557 (\pm 0.105)$, approaching the reliability of experimental assays. Compared to state-of-the-art models, MambaHM achieves a$\mathbf{7. 0 3 3 \%}(\mathbf{\pm 0. 0 4 1})$improvement in cross-cell-type performance and over a 58.978 % (877 bp) improvement in prediction deviation evaluation for peak calling. Overall, MambaHM provides a powerful, cost-effective, and fineresolution tool for histone modification prediction, offering precise epigenomic references for downstream analysis and potential applications in drug discovery. The source code is available at https://github.com/zhichunlizzx/MambaHM.

N6-methyladenine (6mA) is a crucial epigenetic regulator involved in various biological processes. Existing high-throughput methods like 6mA-DIP-seq, ChIP-exo/6mACE-seq, SMRT-seq, and nanopore sequencing offer high accuracy but are typically time-consuming, labor-intensive, and expensive. To address these limitations and promote 6mA research, we introduce XGBoost6mA, a predictive framework integrating deep learning and XGBoost. XGBoost6mA utilizes XGBoost to classify deep learning-derived sequence features, effectively capturing nonlinear patterns and improving classification performance. We evaluated the framework using BERT6mA and CNN6mA feature extractors on benchmark datasets from 11 species. Results showed that XGBoost6mA consistently outperformed both BERT6mA and CNN6mA across all tested species. To validate its robustness, we tested XGBoost6mA by introducing random nucleotide mutations in test sequences. The model maintained stable accuracy, indicating strong resilience to small sequence variations. Additionally, XGBoost6mA offers interpretability by pinpointing key k-mers and positional factors, enhancing the biological significance of its predictions. These findings underscore the advantages of XGBoost6mA in 6mA prediction and its promise as a widely applicable bioinformatics tool. To support future research, XGBoost6mA’s source code is publicly available at https://github.com/xzx0554/xgboost6ma.

Global navigation satellite system-based bistatic synthetic aperture radar interferometry (GNSS-based InBSAR) can improve the monitoring interval to one day due to the using of navigation satellites. Meanwhile, the low signal-to-noise ratio (SNR), poor image resolution, and the random focus position offsets cause large interferometric phase noise. In this letter, an interferometric phase noise reduction algorithm is proposed for GNSS-based InBSAR based on adaptive edge detection and temporal area filtering. An improved edge detection algorithm is adopted to solve the overlapping of resolution cells and phase interference caused by poor resolution. Then, to compensate the random focus position, an area filtering algorithm is proposed to find the temporal supporting area of persistent scatterers (PSs). Finally, the principal phase is extracted to reduce the interferometric phase error. The raw data are used to indicate the effectiveness of the proposed algorithm, and the best monitoring accuracy can reach millimeter level.

The most prominent advantage of the global navigation satellite system-based bistatic synthetic aperture radar interferometry (GNSS-InBSAR) system is its capability to achieve high-frequency 3-D deformation monitoring by combining measurements from different satellites. However, multisatellite, multiband, and short-interval imaging also introduces challenges such as large data amount and high computational requirements. In this article, an adaptive-segmentation-oriented multiband synthesis fast imaging algorithm is proposed for GNSS-InBSAR system. First, multiband synthetic signal model is established for the Beidou navigation signals. Then, optimization methods for segment parameters are proposed along range and azimuth directions, respectively. The limitation of the optimization methods are next discussed based on the GNSS-InBSAR system characteristics. The raw data of multiband navigation signals are used to indicate the effectiveness of the proposed algorithm. The computation time for single-band signals has been reduced by 77%, while for multiband signals, the computation time has been reduced by 71%.

Global Navigation Satellite System-based Bistatic Synthetic Aperture Radar Interferometry (GNSS-InBSAR) system adopts a novel and integrated GNSS receiver, where the direct signal could be collected by the backlobe of the reflected antenna and would result in the fake permanent scatterers (PSs) in the single channel data. Additionally, limited by the data amount, only one channel of data can be transmitted and signal processed. In this letter, an adaptive fake PSs removal algorithm is proposed based on the multi-parameters-estimation direct signal reconstruction. First, the direct signal model considering the transmission link filter, the image offsets, the amplitude, and the phase of the image is established. Then, all these parameters are estimated based on the adaptive image information extraction. Finally, the direct signal is reconstructed to remove the fake PSs in the raw image results. The raw data of the Beidou navigation satellite is used and the final experimental results indicated that the fake PSs in the SAR images can be completely removed, and the deformation retrieval accuracy of the interference area has improved by 63%.

The uncertainty of the elevation of target area in bistatic synthetic aperture radar (BiSAR) introduces image defocus. This uncertainty becomes much worse in global navigation satellite system-based BiSAR interferometry (GNSS-based InBSAR) applications, where the primary problem is a decrease in the coherence of the image pairs. In this paper, a high-coherence oriented imaging algorithm based on adaptive elevation ramp fitting is proposed for GNSS-based InBSAR systems. First, GNSS-based InBSAR signal model is established considering elevation error. From this model, the expressions for position offset and interferometric phase error caused by the elevation error are derived. Then, to improve the elevation fitting accuracy, full-scene fitting is replaced by subarea fitting, and adaptive subarea segmentation is achieved based on the points with complete resolution cells and image valleys. Finally, elevation fitting is performed in the subareas. The algorithm can obtain high-coherence image pairs with low image resolution introduced by GNSS-based InBSAR systems. Simulation and raw data are used to prove the effectiveness of the proposed algorithm in GNSS-based InBSAR.

Pilot contamination incurred by strong inter-cell interference seriously degrades the performance of channel estimation in massive multi-input multi-output (MIMO) orthogonal frequency division multiplexing (OFDM) systems. We propose an independent component analysis (ICA) and sparse recovery algorithm based semi-blind channel estimation scheme, referred to as sparse ICA (SICA), for multi-cell massive MIMO-OFDM systems, which does not require any prior information of intercell interference and therefore is more practical. The proposed SICA scheme enables accurate channel estimation by exploiting both the high-order statistics of the received signal and channel sparsity in angle domain. The SICA scheme performs in a semi-blind manner as it is much more robust against pilot overhead than the previous approaches, and requires only one OFDM symbol as pilot to achieve a superior normalized mean square error of channel estimation. Furthermore, the complexity required by SICA is much lower than that required by the previous work, thanks to the negligible complexity of interference sources number estimation based on sparse recovery algorithm.