Topologically associating domains (TADs) and chromatin architectural loops impact promoter–enhancer interactions, with CCCTC-binding factor (CTCF) defining TAD borders and loop anchors. TAD boundaries and loops progressively strengthen upon embryonic stem (ES) cell differentiation, underscoring the importance of chromatin topology in ontogeny. However, the mechanisms driving this process remain unclear. Here we show a widespread increase in CTCF–RNA-binding protein (RBP) interactions upon ES to neural stem (NS) cell differentiation. While dispensable in ES cells, RBPs reinforce CTCF-anchored chromatin topology in NS cells. We identify Pantr1, a non-coding RNA, as a key facilitator of CTCF–RBP interactions, promoting chromatin maturation. Using acute CTCF degradation, we find that, through its insulator function, CTCF helps maintain neuronal gene silencing in NS cells by acting as a barrier to untimely gene activation during development. Altogether, we reveal a fundamental mechanism driving developmentally linked chromatin structural consolidation and the contribution of this process to the control of gene expression in differentiation.

Motive: Documentation and administration occupy a substantial part of the working time in interventional radiology. Automated drafting tools may assist with these tasks, but their potential benefit requires evaluation. Method: Three predefined simulated cases of peripherally inserted central catheter (PICC) insertion were used to generate reports with ChatGPT on 4 May 2023, using items from the Radiological Society of North America (RSNA) template. Five iterations per case produced 15 report drafts. Eight radiologists, including six board-certified radiologists and two residents, rated completeness, structure, apparent AI authorship, readiness for use without editing and anticipated time savings. Key results: Raters generally assessed completeness and structure positively and anticipated time savings. They generally disagreed that the drafts were clearly identifiable as AI-generated. Some drafts required editing, particularly in relation to report structure and the handling of complications. Documentation time was not measured, and a controlled comparison with human-written reports was not performed. Conclusions: This small simulation study supports the feasibility of AI-assisted PICC report drafting and describes favourable perceptions among the participating raters. It does not establish actual time savings, equivalent report quality or indistinguishability from human-written reports. Prospective evaluation of drafting time, editing requirements and clinical accuracy is needed before conclusions about routine clinical benefit can be drawn.

The rising demand for radiology services calls for innovative solutions to sustain diagnostic quality and efficiency. This study evaluated the diagnostic agreement between two commercially available artificial intelligence (AI) chest X-ray systems and human radiologists during routine clinical practice. We retrospectively analyzed 279 chest X-rays (204 standing, 63 supine, 12 sitting) from a Swiss university hospital. Seven thoracic pathologies – cardiomegaly, consolidation, mediastinal mass, nodule, pleural effusion, pneumothorax, and pulmonary oedema – were assessed. Radiologists’ routine reports were compared against Rayvolve (AZmed) and ChestView (Gleamer, both from Paris, France). A Python code, provided as open access supplement, calculated performance metrics, agreement measures, and effect size quantification. Agreement between radiologists and AI ranged from moderate to almost perfect: Human-AZmed (Gwet’s AC1: 0.47–0.72, moderate to substantial), and Human-Gleamer (Gwet’s AC1: 0.56–0.96, moderate to almost perfect). Balanced accuracies ranged from 0.67–0.85 for Human-AZmed and 0.71–0.85 for Human-Gleamer, with peak performance for pleural effusion (0.85 both systems). Specificity consistently exceeded sensitivity across pathologies (0.70–0.98 vs 0.45–0.85). Common findings showed strong performance, pleural effusion (MCC 0.70–0.73), cardiomegaly (MCC 0.51), and consolidation (MCC 0.45–0.46). Rare pathologies demonstrated lower agreement, mediastinal mass, and nodules (MCC 0.23–0.31). Standing radiographs yielded superior agreement compared to supine studies. The two AI systems showed substantial inter-system agreement for consolidation and pleural effusion (balanced accuracy 0.81–0.84). Both commercial AI chest X-ray systems demonstrated comparable performance to human radiologists for common thoracic pathologies, with no meaningful differences between platforms. Performance was strongest for standing radiographs but declined for rare findings and supine studies. Position-dependent variability and reduced sensitivity for uncommon pathologies underscore the continued need for human oversight in clinical practice.

The demand for improvement of research output evaluation by the international science community has been recently formulated as the Declaration on Research Assessment (DORA) by the American Society for Cell Biology. The Relative Citation Ratio (RCR), introduced by the National Institutes of Health (NIH) is a novel metric indicating the influence of a publication on its peer group — publications from the same research field, as determined by co-citation analysis. The RCR can be viewed for the actual month. While historical RCR data exists within the NIH database, it is not exposed in a way that allows direct or simple access for researchers. We present a method to reconstruct otherwise inaccessible RCR data. A Python-based approach was deployed to extract RCR data from the NIH database and to plot RCR-values for every time point since introduction of the database. This method demonstrates the feasibility of recovering historical bibliometric information and may contribute to more transparent and accountable use of metrics in academic evaluation — in line with the goals of DORA and open science initiatives.

Mutations of genetic sequences are often accompanied by their recombinations, known as phylogenetic networks. These networks are typically reconstructed from coalescent processes that may arise from optimal merging or fitting together a given set of phylogenetic trees. Nakhleh formulated the phylogenetic network reconstruction problem (PNRP): Given a family of phylogenetic trees over a common set of taxa, is there a unique minimal phylogenetic network whose set of spanning trees contains the family? Inspired by ideas from topological data analysis (TDA), we devise lattice-diagram models for phylogenetic networks and filtrations, the cliquegram and the facegram, both generalizing the dendrogram (filtered partition) model of phylogenetic trees. Both models allow us to solve the PNRP rigorously. The solutions are obtained by taking the join of the dendrograms on the lattice of cliquegrams or facegrams. Furthermore, computing the join-facegram is polynomial in the size and number of the input trees. Cliquegrams and facegrams can be challenging to work with when the number of taxa is large. We propose a topological invariant of facegrams and filtrations, called the mergegram, by extending a construction by Elkin and Kurlin defined on dendrograms. We show that the mergegram is invariant of weak equivalences of filtrations which, in turn, implies that it is a 1-Lipschitz stable invariant with respect to Mémoli's tripod distance. The mergegram, can be used as a computable proxy for phylogenetic networks and filtrations of datasets. We illustrate the utility of these new TDA-concepts to phylogenetics, by performing experiments with artificial and biological data.

Motive: Documentation and administration, unpleasant necessities, take a substantial part of the working time in the subspecialty of interventional radiology. With increasing future demand for clinical radiology predicted, time savings from use of text drafting technologies could be a valuable contribution towards our field. Method: Three cases of peripherally inserted central catheter (PICC) line insertion were defined for the present study. The current version of ChatGPT was tasked with drafting reports, following the Radiological Society of North America (RSNA) template. Key results: Score card evaluation by human radiologists indicates that time savings in documentation / administration can be expected without loss of quality from using ChatGPT. Further, automatically generated texts were not assessed to be clearly identifiable as AI-produced. Conclusions: Patients, doctors, and hospital administrators would welcome a reduction of the time that interventional radiologists need for documentation and administration these days. If AI-tools as tested in the present study are brought into clinical application, questions about trust into those systems eg with regard to medical complications will have to be addressed.