Explore the vast range of titles available.

Using a Burkitt lymphoma-like gene expression signature, we recently defined a high-risk molecular high-grade (MHG) group mainly within germinal centre B-cell like diffuse large B-cell lymphomas (GCB-DLBCL), which was enriched for MYC/BCL2 double-hit (MYC/BCL2-DH). The genetic basis underlying MHG-DLBCL and their aggressive clinical behaviour remain unknown. We investigated 697 cases of DLBCL, particularly those with MYC/BCL2-DH (n = 62) by targeted sequencing and gene expression profiling. We showed that DLBCL with MYC/BCL2-DH, and those with BCL2 translocation, harbour the characteristic mutation signatures that are associated with follicular lymphoma and its high-grade transformation. We identified frequent MYC hotspot mutations that affect the phosphorylation site (T58) and its adjacent amino acids, which are important for MYC protein degradation. These MYC mutations were seen in a subset of cases with MYC translocation, but predominantly in those of MHG. The mutations were more frequent in double-hit lymphomas with IG as the MYC translocation partner, and were associated with higher MYC protein expression and poor patient survival. DLBCL with MYC/BCL2-DH and those with BCL2 translocation alone are most likely derived from follicular lymphoma or its precursor lesion, and acquisition of MYC pathogenic mutations may augment MYC function, resulting in aggressive clinical behaviour.

Despite the well-established significance of transcription factors (TFs) in pathogenesis, their utilization as pharmacological targets has been limited by the inherent challenges in modulating their protein interactions. The lack of defined small-molecule binding pockets and the nuclear localization of TFs do not favor the use of traditional tools. Aptamers possess large molecular weights, expansive blocking surfaces and efficient cellular internalization, making them compelling tools for modulating TF interactions. Here, we report a structure-guided design strategy called Blocker-SELEX to develop inhibitory aptamers (iAptamers) that selectively block TF interactions. Our approach leads to the discovery of iAptamers that cooperatively disrupt SCAF4/SCAF8-RNAP2 interactions, dysregulating RNAP2-dependent gene expression, which impairs cell proliferation. This approach is further applied to develop iAptamers blocking WDR5-MYC interactions. Overall, our study highlights the potential of iAptamers in disrupting pathogenic TF interactions, implicating their potential utility in studying the biological functions of TF interactions and in nucleic acids drug discovery. Transcription factors are crucial in disease but hard to target with traditional drugs. Here, authors present BlockerSELEX, a strategy to develop inhibitory aptamers that block transcription factor interactions, which disrupts interactions between key proteins, showing potential for new nucleic acid therapies.

Artificial intelligence (AI) is opening new frontiers in the development of antibody-drug conjugates (ADCs), offering unprecedented opportunities for precision therapy. This review outlines how AI empowers each stage of the ADC pipeline. In target discovery, multi-omics integration and graph-based learning prioritize tumor-selective and internalizing antigens. In antibody engineering, structure prediction, affinity optimization, and developability modeling streamline candidate selection. For linker-payload design, generative models and multi-objective optimization approaches support the rational design of conjugates that balance potency, stability, and immunogenicity. In absorption, distribution, metabolism, excretion, and toxicity (ADMET) modeling, deep learning and transformer-based frameworks predict pharmacokinetics and toxicity with increasing accuracy and mechanistic clarity. In clinical development, AI facilitates patient stratification, response prediction, and trial simulation through digital twin models, adaptive dosing algorithms, and real-world data integration. These capabilities support a more personalized and efficient pathway from bench to bedside. To further realize the impact of AI in ADC development, we highlight strategic priorities including the creation of curated, multimodal datasets, interpretable model architectures, and closed-loop experimental platforms. Together, these advances will be essential for realizing the full potential of AI to support rational, scalable, and personalized ADC-based therapies in oncology.

Accurate preoperative prediction of cervical Lymph Node Metastasis (LNM) is critical for surgical decision-making in thyroid cancer patients, and the difficulty in it often leads to over-treatment. UltraSound (US) and Computed Tomography (CT) are two primary non-invasive examinations, but neither method alone provides satisfactory diagnostic accuracy. To address this problem, we propose a Multimodal Nested Attention Network (MNANet) to integrate US and CT images. The network is designed to extract specific complementary information from US and CT images, and comprehensively fuse multimodal features at multiple granularities. In our internal cohort, MNANet achieves Areas Under the Curves (AUCs) of 0.88 and 0.86 for central and lateral cervical sites, respectively, representing a significant improvement of 0.06 to 0.10 compared to unimodal models and outperforming state-of-the-art medical multimodal methods across all metrics.The model demonstrates robust cross-institutional generalization and maintains superior performance across other imaging modalities(e.g., Magnetic Resonance Imaging (MRI)). Additionally, our model exhibits a more precise focus on the thyroid nodule, indicating enhanced learning ability. Moreover, we systematically evaluate the applicability across various clinical characteristics, identifying individuals who can benefit most from the multimodal approach.

Background Classifiers based on molecular criteria such as gene expression signatures have been developed to distinguish Burkitt lymphoma and diffuse large B cell lymphoma, which help to explore the intermediate cases where traditional diagnosis is difficult. Transfer of these research classifiers into a clinical setting is challenging because there are competing classifiers in the literature based on different methodology and gene sets with no clear best choice; classifiers based on one expression measurement platform may not transfer effectively to another; and, classifiers developed using fresh frozen samples may not work effectively with the commonly used and more convenient formalin fixed paraffin-embedded samples used in routine diagnosis. Methods Here we thoroughly compared two published high profile classifiers developed on data from different Affymetrix array platforms and fresh-frozen tissue, examining their transferability and concordance. Based on this analysis, a new Burkitt and diffuse large B cell lymphoma classifier (BDC) was developed and employed on Illumina DASL data from our own paraffin-embedded samples, allowing comparison with the diagnosis made in a central haematopathology laboratory and evaluation of clinical relevance. Results We show that both previous classifiers can be recapitulated using very much smaller gene sets than originally employed, and that the classification result is closely dependent on the Burkitt lymphoma criteria applied in the training set. The BDC classification on our data exhibits high agreement (~95 %) with the original diagnosis. A simple outcome comparison in the patients presenting intermediate features on conventional criteria suggests that the cases classified as Burkitt lymphoma by BDC have worse response to standard diffuse large B cell lymphoma treatment than those classified as diffuse large B cell lymphoma. Conclusions In this study, we comprehensively investigate two previous Burkitt lymphoma molecular classifiers, and implement a new gene expression classifier, BDC, that works effectively on paraffin-embedded samples and provides useful information for treatment decisions. The classifier is available as a free software package under the GNU public licence within the R statistical software environment through the link http://www.bioinformatics.leeds.ac.uk/labpages/softwares/ or on github https://github.com/Sharlene/BDC .

Biologically distinct subtypes of diffuse large B-cell lymphoma can be identified using gene-expression analysis to determine their cell of origin, corresponding to germinal centre or activated B cell. We aimed to investigate whether adding bortezomib to standard therapy could improve outcomes in patients with these subtypes. In a randomised evaluation of molecular guided therapy for diffuse large B-cell lymphoma with bortezomib (REMoDL-B), an open-label, adaptive, randomised controlled, phase 3 superiority trial, participants were recruited from 107 cancer centres in the UK (n=94) and Switzerland (n=13). Eligible patients had previously untreated, histologically confirmed diffuse large B-cell lymphoma with sufficient diagnostic material from initial biopsies for gene-expression profiling and pathology review; were aged 18 years or older; had ECOG performance status of 2 or less; had bulky stage I or stage II–IV disease requiring full-course chemotherapy; had measurable disease; and had cardiac, lung, renal, and liver function sufficient to tolerate chemotherapy. Patients initially received one 21-day cycle of standard rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisolone (R-CHOP; rituximab 375 mg/m2, cyclophosphamide 750 mg/m2, doxorubicin 50 mg/m2, and vincristine 1·4 mg/m2 [to a maximum of 2 mg total dose] intravenously on day 1 of the cycle, and prednisolone 100 mg orally once daily on days 1–5). During this time, we did gene-expression profiling using whole genome cDNA-mediated annealing, selection, extension, and ligation assay of tissue from routine diagnostic biopsy samples to determine the cell-of-origin subtype of each participant (germinal centre B cell, activated B cell, or unclassified). Patients were then centrally randomly assigned (1:1) via a web-based system, with block randomisation stratified by international prognostic index score and cell-of-origin subtype, to continue R-CHOP alone (R-CHOP group; control), or with bortezomib (RB-CHOP group; experimental; 1·3 mg/m2 intravenously or 1·6 mg/m2 subcutaneously) on days 1 and 8 for cycles two to six. If RNA extracted from the diagnostic tissues was of insufficient quality or quantity, participants were given R-CHOP as per the control group. The primary endpoint was 30-month progression-free survival, for the germinal centre and activated B-cell population. The primary analysis was on the modified intention-to-treat population of activated and germinal centre B-cell population. Safety was assessed in all participants who were given at least one dose of study drug. We report the progression-free survival and safety outcomes for patients in the follow-up phase after the required number of events occurred. This study was registered at ClinicalTrials.gov, number NCT01324596, and recruitment and treatment has completed for all participants, with long-term follow-up ongoing. Between June 2, 2011, and June 10, 2015, 1128 eligible patients were registered, of whom 918 (81%) were randomly assigned to receive treatment (n=459 to R-CHOP, n=459 to RB-CHOP), comprising 244 (26·6%) with activated B-cell disease, 475 (51·7%) with germinal centre B cell disease, and 199 (21·7%) with unclassified disease. At a median follow-up of 29·7 months (95% CI 29·0–32·0), we saw no evidence for a difference in progression-free survival in the combined germinal centre and activated B-cell population between R-CHOP and RB-CHOP (30-month progression-free survival 70·1%, 95% CI 65·0–74·7 vs 74·3%, 69·3–78·7; hazard ratio 0·86, 95% CI 0·65–1·13; p=0·28). The most common grade 3 or worse adverse event was haematological toxicity, reported in 178 (39·8%) of 447 patients given R-CHOP and 187 (42·1%) of 444 given RB-CHOP. However, RB-CHOP was not associated with increased haematological toxicity and 398 [87·1%] of 459 participants assigned to receive RB-CHOP completed six cycles of treatment. Grade 3 or worse neuropathy occurred in 17 (3·8%) patients given RB-CHOP versus eight (1·8%) given R-CHOP. Serious adverse events occurred in 190 (42·5%) patients given R-CHOP, including five treatment-related deaths, and 223 (50·2%) given RB-CHOP, including four treatment-related deaths. This is the first large-scale study in diffuse large B-cell lymphoma to use real-time molecular characterisation for prospective stratification, randomisation, and subsequent analysis of biologically distinct subgroups of patients. The addition of bortezomib did not improve progression-free survival. Janssen-Cilag, Bloodwise, and Cancer Research UK.

Burkitt lymphoma and diffuse large B-cell lymphoma are two closely related types of lymphoma that are managed differently in clinical practice and the accurate diagnosis is a key point in treatment decisions. However based on current criteria combined with morphological, immunophenotypic and genetic characteristics, a significant number of cases exhibit overlapping features where diagnosis and treatment decisions are difficult to make. Especially, the prognosis have been reported significantly unfavourable in a subset of cases that are initially diagnosed as diffuse large B-cell lymphoma but bear MYC gene translocation, which is a defining feature of Burkitt lymphoma however can also be found in other lymphomas. Despite the adverse effect of MYC in aggressive lymphomas other than Burkitt lymphoma, the underlying mechanism and effective treatment is still unclear. Recent technological advances have made it possible to simultaneously investigate an enormous number of bio-molecules, and the scientific fields associated with measuring molecular data in such a high-throughput way are usually called “omics”. For example, genomics assesses thousands of DNA sequences and transcriptomics assays large numbers of transcripts in a single experiment. These techniques together with the rapidly emerging analytical methods in bioinformatics have introduced cancer research into a new era. The growing amount of omics data have significantly influenced the understanding of lymphomas and hold great promise in classifying subtypes, predicting treatment responses that will eventually lead to personalized therapy. Here in this study, we investigate the discrimination of Burkitt lymphoma and diffuse large B-cell lymphoma based on DNA microarray gene expression data, which has contributed most in molecular classification of lymphoma subtypes in the last decade. On the basis of two previous research level gene expression profiling classifiers, we developed a robust classifier that works effectively on different platforms and formalin fixed paraffin-embedded samples commonly used in routine clinic. The validation of the classifier on the samples from clinical patients achieves a high agreement with diagnosis made in a central haematopathology laboratory, and leads to a potential outcome indication in the patients presenting intermediate features. In addition, we explore the role of MYC in the above lymphomas. Our investigation emphasizes the inferior impact of high level MYC mRNA expression on patients’ outcome, and the functional analysis of MYC high expression associated genes show significantly enriched molecular mechanisms of proliferation and metabolic process. Moreover, the gene PRMT5 is found to be highly correlated with MYC expression which opens a possible therapeutic target for the treatment.

Although substantial efforts have been made using graph neural networks (GNNs) for AI-driven drug discovery (AIDD), effective molecular representation learning remains an open challenge, especially in the case of insufficient labeled molecules. Recent studies suggest that big GNN models pre-trained by self-supervised learning on unlabeled datasets enable better transfer performance in downstream molecular property prediction tasks. However, the approaches in these studies require multiple complex self-supervised tasks and large-scale datasets, which are time-consuming, computationally expensive, and difficult to pre-train end-to-end. Here, we design a simple yet effective self-supervised strategy to simultaneously learn local and global information about molecules, and further propose a novel bi-branch masked graph transformer autoencoder (BatmanNet) to learn molecular representations. BatmanNet features two tailored complementary and asymmetric graph autoencoders to reconstruct the missing nodes and edges, respectively, from a masked molecular graph. With this design, BatmanNet can effectively capture the underlying structure and semantic information of molecules, thus improving the performance of molecular representation. BatmanNet achieves state-of-the-art results for multiple drug discovery tasks, including molecular properties prediction, drug-drug interaction, and drug-target interaction, on 13 benchmark datasets, demonstrating its great potential and superiority in molecular representation learning.

Advances in single-cell multimodal profiling have enabled a more systematic analysis of cellular biology, yet the rapid accumulation of large-scale, heterogeneous datasets poses substantial challenges for integrative analysis. Recently, Transformer-based cell language models (CLMs) are becoming powerful foundational tools for learning transferable cell representations from unimodal single-cell datasets. However, a flexible and unified multimodal foundation models for joint modeling of scRNA-seq and scATAC-seq datasets remains lacking. Here, we present CLM-X, a multimodal single-cell foundation model built on multiway Transformer architecture. CLM-X employs a harmonized tokenization design together with a stage-wise masked reconstruction pretraining strategy, enabling unified modeling of RNA-only, ATAC-only, and paired RNA-ATAC-paired input within a single Transformer-based framework. We pretrain CLM-X on million-scale unimodal and multimodal datasets, and systematically evaluate its transferability on five downstream tasks including batch correction, modality integration, cross-modal translation, cell type annotation, and perturbation prediction. Across comprehensive benchmarks on 10 datasets, CLM-X consistently outperforms existing multimodal methods and unimodal foundation models, with particularly clear advantages in RNA-ATAC cross-modal translation and genetic-perturbation-response prediction. Overall, CLM-X establishes a unified and scalable multimodal foundation model for integrative analysis of scRNA-seq and scATAC-seq datasets, advancing a more robust, comprehensive, and biological interpretable single-cell analysis beyond task-specific approaches and unimodal foundation models.Competing Interest StatementThe authors have declared no competing interest.