Explore the vast range of titles available.

Background Thalassemia, a hereditary hemolytic blood disorder, is characterized by high global allele prevalence and a gradually expanding variant spectrum. In Guangxi, China, routine genetic screening targeting hotspot variants of thalassemia has been implemented for over a decade, significantly reducing the incidence of severe cases. This study assessed the efficacy of the existing thalassemia screening protocol in Guangxi, China, by applying long-read sequencing (LRS) in family trios/quartets analysis for newborns. Methods This study enrolled 1491 families for the family trios/quartets analysis, among which at least one of the parents had undergone routine genetic screening for thalassemia. A total of 1516 newborns from these families were tested using LRS. The efficacy of routine screening was assessed by comparing detection results across generations. Results α-thalassemia demonstrated higher prevalence than β-thalassemia among newborns, with -- SEA and HBB c.52A > T as the predominant variants. Among the 1516 newborns, 1348 (88.92%) exhibited LRS results that were concordant with the predictions from parental screening. Of these 1348 newborns, 28 had parents who had undergone routine prenatal diagnosis. Discordance between the predictions from parental screening and newborn test results occurred in 168 cases (11.08%), with 154 (10.16%) of these attributable to genetic testing of only one of the parents or incomplete testing of both parents, while the remaining 14 (0.92%) were attributed to limitations of routine methods in detecting complex structural variants and rare point mutations. Conclusions Although routine genetic screening targeting hotspot variants is highly effective, its precision is constrained by strategic and methodological limitations. A comprehensive and accurate LRS-based screening method can serve as a valuable supplement to routine screening, enhancing the precision of thalassemia prevention and management.

Background Deletion or duplication in the DMD gene is one of the most common causes of Duchenne and Becker muscular dystrophy (DMD/BMD). However, the pathogenicity of complex rearrangements involving DMD , especially segmental duplications with unknown breakpoints, is not well understood. This study aimed to evaluate the structure, pattern, and potential impact of rearrangements involving DMD duplication. Methods Two families with DMD segmental duplications exhibiting phenotypical differences were recruited. Optical genome mapping (OGM) was used to explore the cryptic pattern of the rearrangements. Breakpoints were validated using long-range polymerase chain reaction combined with next-generation sequencing and Sanger sequencing. Results A multi-copy duplication involving exons 64–79 of DMD was identified in Family A without obvious clinical symptoms. Family B exhibited typical DMD neuromuscular manifestations and presented a duplication involving exons 10–13 of DMD . The rearrangement in Family A involved complex in-cis tandem repeats shown by OGM but retained a complete copy (reading frame) of DMD inferred from breakpoint validation. A reversed insertion with a segmental repeat was identified in Family B by OGM, which was predicted to disrupt the normal structure and reading frame of DMD after confirming the breakpoints. Conclusions Validating breakpoint and rearrangement pattern is crucial for the functional annotation and pathogenic classification of genomic structural variations. OGM provides valuable insights into etiological analysis of DMD/BMD and enhances our understanding for cryptic effects of complex rearrangements.

Background Disorders of sex development (DSDs) exhibit high genetic and phenotypic heterogeneity, and genotype–phenotype correlations are not fully understood. 5α-Reductase type 2 (5α-RD2) deficiency, a common form of DSD, is caused by SRD5A2 inactivation. This study investigated the role of SRD5A2 haplotypes in DSD, focusing on their corresponding phenotypes, structural changes and impacts on enzyme activity. Methods This study enrolled 216 individuals with DSD who underwent genetic analysis and 2,794 controls. Linkage disequilibrium analysis was performed in individuals with 5α-RD2 deficiency to identify SRD5A2 haplotypes, and haplotype frequencies were analysed across cohorts. The clinical manifestations of individuals with different SRD5A2 haplotypes were characterized. Structural predictions were employed to investigate the impacts of haplotypes on the 5α-RD2 structure and interactions with ligands. Functionally, kinetic assays were conducted to validate the effects of different haplotypes on enzyme activity. Results A SRD5A2 haplotype composed of c.265C > G and c.680G > A (Hap3: G-A) was identified, and the haplotype frequency was 64.71% in individuals with 5α-RD2 deficiency, 2.59% and 1.22% in non-5α-RD2 deficiency DSD cases without or with known DSD-related gene variants, respectively, and 1.57% in in-house controls. Globally, Hap3: G-A was enriched in southern Chinese individuals and showed high population differentiation, indicating a potential founder effect of the haplotype. The majority of homozygotes of Hap3: G-A presented microphallus, and nearly half of them manifested isolated microphallus. Structurally, Hap3: G-A was predicted to result in an increase in the solvent-accessible surface area (10.72 Å 2 ), a redistribution of hydrogen bonds within 5α-RD2, and a loss of key hydrogen bonds with NADPH. Functionally, kinetic assays showed that the catalytic efficiency of the enzyme encoded by Hap3: G-A was between that of Hap1: G-G and that of Hap2: C-A. Conclusions Hap3: G-A, which is prevalent in individuals with 5α-RD2 deficiency, suggests a potential founder effect. Structurally, compared with other haplotypes, Hap3: G-A seems to have a combined effect on the structure and interaction of 5α-RD2, rather than have merely additive effects of its constituent variants. Functionally, kinetic assays suggested a hypomorphic effect of Hap3: G-A. These findings provide valuable insights for understanding genotype–phenotype correlations, genetic counselling, early intervention and clinical management of individuals with 5α-RD2 deficiency or even other DSDs. Highlights On the basis of 216 DSD individuals and 2,794 controls, a novel SRD5A2 haplotype (Hap3: G-A), which is composed of common variants and is especially prevalent in DSD patients with 5α-RD2 deficiency, was identified, indicating the potential founder effect of Hap3: G-A. The majority of 5α-RD2 deficiency individuals with homozygous Hap3: G-A presented microphallus. Three-dimensional structure and model construction predicted that Hap3: G-A resulted in an increase in the solvent-accessible surface area, redistribution of hydrogen bonds within 5α-RD2, and loss of key hydrogen bonds with NADPH. Functionally, kinetic assays showed that the catalytic efficiency of the enzyme encoded by Hap3: G-A was between those of Hap1: G-G and Hap2: C-A, suggesting a hypomorphic effect of this haplotype. Plain Language Summary Individuals with disorders of sex development (DSDs) present variable genotypes and phenotypes, and the genotype–phenotype correlation remains poorly understood. 5α-Reductase type 2 (5α-RD2) deficiency, a common form of DSD, is caused by SRD5A2 inactivation. In this study, genetic testing and analysis were performed in 216 individuals with DSD and 2,794 non-DSD controls. A novel SRD5A2 haplotype (Hap3: G-A) was identified, and the haplotype frequency was 64.71% in individuals with 5α-RD2 deficiency, 2.59% and 1.22% in non-5α-RD2 deficiency DSD cases without or with known DSD-related gene variants, respectively, and 1.57% in in-house controls. Globally, Hap3: G-A was enriched in southern Chinese individuals and showed high population differentiation, indicating a potential founder effect of the haplotype. Analyzing the phenotype spectrum of individuals with 5α-RD2 deficiency, we found that the majority of homozygotes of Hap3: G-A presented microphallus, and nearly half of them manifested isolated microphallus. Structurally, Hap3: G-A was predicted to result in an increase in the solvent-accessible surface area, the redistribution of hydrogen bonds within 5α-RD2, and the loss of key hydrogen bonds with NADPH. Compared with other haplotypes, Hap3: G-A seemed to have a combined effect on the structure and interaction of 5α-RD2, rather than simply additive effects of its constituent variants. Functionally, kinetic assays showed that the catalytic efficiency of the enzyme encoded by Hap3: G-A was between that of Hap1: G-G and that of Hap2: C-A, suggesting a hypomorphic effect of this haplotype. These findings provide valuable insights for understanding genotype–phenotype correlations, genetic counselling, early intervention and clinical management of individuals with 5α-RD2 deficiency or even other DSDs.

In this study, we elucidated the exponential synchronization of a complex network system with time-varying delay. Then the exponential synchronization control of several types of complex network systems with time-varying delay under no requirements of delay derivable were explored. The dynamic behavior of a system node shows time-varying delays. Thus, to derive suitable conditions for the exponential synchronization of different complex network systems, we designed a linear feedback controller for linear coupling functions, using the Lyapunov stability theory, Razumikhin theorem, and Newton–Leibniz formula. The exponential damping rates for the exponential synchronization of different complex network systems were then estimated. Finally, we validated our conclusions through a numerical simulation.

In this study, a novel auto-tuned hybrid deep learning approach composed of three base deep learning models, namely, long short-term memory, gated recurrent unit, and support vector regression, is developed to predict the fracture evolution process. The novelty of this framework lies in the auto-determined hyperparameter configurations for each base model based on the Bayesian optimization technique, which guarantees the fast and easy implementation in various practical applications. Moreover, the ensemble modeling technique auto consolidates the predictive capability of each base model to generate the final optimized hybrid model, which offers a better prediction of the overall fracture pattern evolution, as demonstrated by a case study. The comparison of the different prediction strategies exhibits that the direct prediction is a better option than the recursive prediction, in particular for a longer prediction distance. The proposed approach may be applied in various sequential data predictions by adopting the adaptive prediction scheme.

Fine-grained power data are used to characterize user's behavior in smart grid, Non-intrusive load monitoring can effectively separate the load of a single electrical appliance from the whole energy consumption of a dwelling, which is beneficial to fully exploit the load potential. In this paper, considering the relationship on usage habits among electrical appliances, the characteristics of energy consumption, we propose a multi-objective modeling method based on deep learning. The multi-objective model is constructed by CNN and LSTM, and the neural network is collaboratively optimized by multi-objective outputs. The experimental results show that the multi-objective model performs better than the other models on the five appliances, reducing the absolute error to less than 10 and the total error of the standardized signal to less than 0.1.

Concept erasure aims to prevent image generative models from producing unsafe content while preserving their general generative capability. Meanwhile, next-scale autoregressive (AR) image generation has recently emerged as a new generative paradigm characterized by next-scale prediction, for which concept erasure remains largely unexplored. In this paradigm, semantic information is highly compressed at early scales, leading to severe entanglement between unsafe and unrelated semantics. In this paper, we propose ScaleErasure, an inference-time concept erasure method that performs minimal intervention. ScaleErasure precisely selects and guides predicted logits that are most relevant to the unsafe concept, thereby enabling effective erasure under severe semantic entanglement. Specifically, ScaleErasure performs two additional forward passes conditioned on the unsafe concept and the corresponding safe concept, and leverages their outputs to guide the target logits away from unsafe concepts toward safe concepts. To enable precise and minimal intervention, logits selection and guidance are conducted across three dimensions: scales, tokens, and bit channels. Experiments demonstrate that ScaleErasure outperforms adapted baselines in the next-scale AR paradigm, achieving more precise concept erasure while largely preserving general generative capability. The code is available at https://github.com/coziiizz/ScaleErasure.

Multi-label recognition with frozen Vision-Language Models (VLMs) is brittle under distribution shift: standard zero-shot inference scores labels independently, ignoring co-occurrence structure and producing incoherent label sets where dominant concepts suppress weaker but compatible labels. We introduce Bayesian Conditional Priors (BCP) Estimation, a gradient-free test-time adaptation method that injects label dependency without tuning the backbone. BCP views zero-shot logits as a proxy for marginal posteriors under a fixed image-text likelihood and attributes shift-induced errors mainly to a mismatched label prior. For each test image, it selects a high-confidence anchor label and applies an anchor-conditioned Bayesian refinement. This update is closed-form in logit space and admits a pointwise mutual information (PMI) interpretation, explicitly promoting compatible labels and suppressing incompatible ones. BCP operates without target annotations by estimating anchor-conditioned priors online from the unlabeled test stream via lightweight second-order co-occurrence statistics, adding negligible overhead beyond a single forward pass. Across standard multi-label benchmarks and multiple CLIP backbones, BCP consistently outperforms strong TTA baselines, e.g., improving RN50 average mAP from 57.31 to 69.22 and ViT-B/16 from 62.61 to 71.79.

Embedding-as-a-Service (EaaS) is an effective and convenient deployment solution for addressing various NLP tasks. Nevertheless, recent research has shown that EaaS is vulnerable to model extraction attacks, which could lead to significant economic losses for model providers. For copyright protection, existing methods inject watermark embeddings into text embeddings and use them to detect copyright infringement. However, current watermarking methods often resist only a subset of attacks and fail to provide comprehensive protection. To this end, we present the region-triggered semantic watermarking framework called RegionMarker, which defines trigger regions within a low-dimensional space and injects watermarks into text embeddings associated with these regions. By utilizing a secret dimensionality reduction matrix to project onto this subspace and randomly selecting trigger regions, RegionMarker makes it difficult for watermark removal attacks to evade detection. Furthermore, by embedding watermarks across the entire trigger region and using the text embedding as the watermark, RegionMarker is resilient to both paraphrasing and dimension-perturbation attacks. Extensive experiments on various datasets show that RegionMarker is effective in resisting different attack methods, thereby protecting the copyright of EaaS.

Large language models (LLMs) have achieved remarkable performance across many generation tasks. Nevertheless, effectively aligning them with desired behaviors remains a significant challenge. Activation steering is an effective and cost-efficient approach that directly modifies the activations of LLMs during the inference stage, aligning their responses with the desired behaviors and avoiding the high cost of fine-tuning. Existing methods typically indiscriminately intervene to all generations or rely solely on the question to determine intervention, which limits the accurate assessment of the intervention strength. To this end, we propose the Flexible Activation Steering with Backtracking (FASB) framework, which dynamically determines both the necessity and strength of intervention by tracking the internal states of the LLMs during generation, considering both the question and the generated content. Since intervening after detecting a deviation from the desired behavior is often too late, we further propose the backtracking mechanism to correct the deviated tokens and steer the LLMs toward the desired behavior. Extensive experiments on the TruthfulQA dataset and six multiple-choice datasets demonstrate that our method outperforms baselines. Our code will be released at https://github.com/gjw185/FASB.