Narcissism is a hot topic in popular media, which often emphasizes harmful aspects and behaviours of individuals with narcissistic features. Such media likely contributes to negative perspectives that could further stigmatize narcissism and those who may suffer from narcissistic difficulties. However, limited research has investigated perspectives about narcissism among the general public. The present study surveyed 815 US-based adults to explore public perspectives regarding the prevalence of narcissism, attitudes related to stigma, perspectives on suffering and hope, therapeutic issues and experiences of harm associated with narcissism. Findings indicate that a majority view narcissism as a prevalent and undesirable personality feature and warrant considerable clinical intervention. Individuals who endorsed having experienced harm by someone with narcissistic features were more likely to express negative perspectives. The mixture of public perspectives regarding narcissism seems to indicate a potential role for public education in order to reduce stigma and promote help-seeking and treatment efforts.

Radiomics is increasingly explored as a tool for improving diagnosis, prognosis, and treatment planning. However, concerns exist about the reproducibility and methodological rigor of its studies. The integration of high-dimensional radiomic features and machine learning makes the field prone to unintentional errors that may warrant retraction. Despite a rising number of retractions in science overall, no dedicated study has examined retractions specifically within radiomics. Therefore, this study aimed to review retracted radiomics publications and identify the characteristics and reasons for their retraction. We systematically searched sixdatabases (Crossref, Retraction Watch Database, OpenAlex, PubMed, Scopus, Web of Science) and identified 93 retracted radiomics publications, of which 20 were included. These articles were analyzed with respect to publisher, country of origin, dates, citation counts, and reasons for retraction. Retraction rates were then estimated and compared with those in general radiology. Our findings indicate that a disproportionate number of retractions are linked to specific publishers and countries (particularly China and India), with overall low citation counts (median 4.0 citations). Retractions peaked sharply in 2023, followed by a strong decline. Many retraction notes lack a clear explanation for the retraction. Estimated retraction rates in radiomics were lower than in general radiology (6.7 vs 7.4 per 10,000 publications). Notably, no major radiological or oncological journal appears to have retracted a radiomics publication. Given that radiomics demands higher, interdisciplinary expertise, this suggests a gap, implying that flawed research may yet have to be retracted. KEY POINTS: Question Considering the technical complexity of radiomics studies and their susceptibility to unintentional errors, how do their retraction rates compare to those in general radiology? Findings Retractions in radiomics were disproportionately linked to specific publishers and countries; however, no retractions appeared in major journals. Estimated retraction rates were lower than those for general radiology publications. Clinical relevance A potential gap in the number of retracted radiomics studies was identified, implying that flawed research in the field may not yethave been addressed.

Abstract We introduce GPUmonty , a CUDA/C-based Monte Carlo radiative transfer code accelerated using graphics processing units (GPUs). GPUmonty derives from the CPU-based code grmonty and offloads the most computationally expensive stages of the calculation—superphoton generation, sampling, tracking, and scattering—to the GPU. Whereas grmonty handles photons sequentially, GPUmonty processes large numbers of superphotons concurrently, leveraging the single-instruction, multiple-thread execution model of modern GPUs. Benchmarks demonstrate a speedup of about 12× relative to the original CPU implementation on a single GPU, with runtime limited primarily by register pressure rather than compute or memory bandwidth saturation. We validate the implementation through analytic tests for a optically thin synchrotron sphere, as well as comparisons with igrmonty for scattering synchrotron sphere and general relativistic magnetohydrodynamic simulation data. Relative errors remain below a percent level, and convergence is consistent with the expected N s − 1 / 2 Monte Carlo scaling. By significantly reducing computational costs, GPUmonty enables the extensive parameter space surveys and faster spectra modeling required to interpret horizon-scale observations of supermassive black holes. GPUmonty is publicly available under the GNU (GNU's Not Unix) General Public License.

Mitochondrial transcript abundance is a standard quality control metric in single-cell RNA sequencing, but fixed percentage thresholds fail to account for the substantial variation in mitochondrial content across cell types and tissues, risking both retention of compromised cells and exclusion of transcriptionally active viable cell populations. We present MitoChontrol, a cell-type-aware probabilistic framework for mitochondrial quality control that models the mitochondrial transcript fraction within transcriptionally coherent clusters as a Gaussian mixture distribution. Compromised-cell components are identified from the upper tail of each cluster-specific distribution, and filtering thresholds are defined as the point at which the posterior probability of cellular compromise exceeds a user-definded confidence value. Applied to controlled perturbation experiments and a pancreatic ductal adenocarcinoma single-cell dataset, MitoChontrol selectively removes transcriptionally compromised cells while preserving biologically elevated but viable populations, outperforming fixed-threshold and outlier-based approaches. MitoChontrol is implemented in Python and integrates directly with AnnData-based workflows. It is freely available under the GNU General Public License v3 (GPL-3.0) at: https://github.com/uttamLab/MitoChontrol (DOI: https://doi.org/10.5281/zenodo.19423054 ).

High-throughput environmental DNA sequencing has ushered ecological and evolutionary studies into the big data era. With thousands to millions of DNA sequences, designing taxon-specific oligonucleotides is a current bottleneck of molecular studies that rely on primers for Polymerase Chain Reactions (PCRs) or probes for Fluorescence insitu Hybridization (FISH). No software currently exists to design specific oligonucleotides starting from a custom set of sequences. Existing tools rely on specific databases, alignments or phylogenetic trees, or cannot accommodate increasingly large molecular environmental datasets. Here we present oligoN-design, a versatile tool to design oligonucleotides specific to a set of target sequences while minimizing predicted binding to non-target sequences. OligoN-design is simple, reproducible and adaptable to high-throughput sequencing data analyses. It requires only two fasta files as input, one containing target taxa and the other containing non-target taxa. Using standard bioinformatic formats, it integrates easily with other tools such as BLAST, VSEARCH or MAFFT. OligoN-design allows a range of strategies that we present in detail, from an unsupervised end-to-end usage all the way to a detailed and thorough expert usage. Starting with large, comprehensive ribosomal databases that are widely used by the community (i.e., PR2, SILVA) and the unsupervised function, we were able to replicate known taxa-specific oligonucleotides in under 30 min and up to 6 GB of RAM on a personal laptop. OligoN-design, available at github.com/MiguelMSandin/oligoN-design under GNU General Public Licence version 3.0, is easily installed via bioconda bioconda.github.io/recipes/oligon-design/README.html.

Liquid biopsy offers a non-invasive approach to study tumor-derived genetic material circulating in plasma. Beyond genetic alterations, the fragmentomic features of cell-free DNA-such as fragment size, genomic position, and end-motifs-provide valuable insights into the biological and clinical context of DNA release. fRagmentomics is a user-friendly R package designed to characterize cfDNA fragments overlapping one or multiple small mutations of any type, starting from an aligned sequencing file (BAM). It supports multiple mutation input formats, accommodates one-based and zero-based genomic conventions, resolves mutation representation ambiguities, and accepts any reference file in FASTA format. For each fragment overlapping a mutation of interest, fRagmentomics outputs fragment-level features including its fragment size, end-motifs, and mutational status, along with additional fragment-level or read-level information. The package implements an indel-aware and optionally soft-clip-preserving fragment size computation that improves accuracy over conventional size estimates based solely on aligned positions. fRagmentomics is licensed under GNU General Public License v3.0 and available at https://github.com/ElsaB-Lab/fRagmentomics, https://anaconda.org/elsab-lab/r-fragmentomics and https://bioconductor.org/packages/fRagmentomics, with documentation and a tutorial. yoann.pradat@gustaveroussy.fr, elsa.bernard@gustaveroussy.fr. Supplementary data are available at Bioinformatics online.

Lung cancer remains the leading cause of cancer-related mortality globally and is often diagnosed at advanced stages in Saudi Arabia. This cross-sectional study aimed to quantify public awareness and knowledge of lung cancer screening (LCS) using LDCT and identify barriers to its implementation in Riyadh. A validated 24-item questionnaire was administered to 452 participants to assess demographic factors, smoking history, and LCS knowledge. Results revealed that only 30.1% of participants had heard of LCS, and 50.2% demonstrated "poor" knowledge scores (mean score 11.0 ± 4.97). Higher knowledge scores were significantly associated with being female, having a bachelor's degree or higher, and being a non-smoker. While 78.1% expressed willingness to undergo screening, the most significant barrier was a lack of knowledge about the test (44.1%), followed by concerns regarding radiation exposure (36.1%). Conversely, a healthcare provider's recommendation was identified as the primary motivator for 53.3% of respondents. These findings highlight a critical "awareness-willingness" gap. While public willingness is high, this should not be misconstrued as systemic preparedness; substantial educational and structural gaps remain that must be bridged before national implementation can be considered feasible. We conclude that while public willingness is high, successful implementation requires a transition toward organized invitation systems and the use of multifactorial risk profiles. Integrating epidemiological evidence with proactive policy design is essential to ensure that the national program avoids systematic under- or over-inclusion and remains effective for all demographics.

Inappropriate use of medicines remains a widespread issue encountered in daily life. Furthermore, there is a general reluctance to engage in proactive behaviors related to medicine management, such as independently maintaining a medication notebook. To address these challenges, we developed and implemented an interactive and engaging learning program targeted at the general public to promote behavioral and attitudinal changes toward the responsible use of pharmaceuticals. The program was based on the attention, relevance, confidence, and satisfaction (ARCS) model—an educational framework that emphasizes attention, relevance, confidence, and satisfaction—and additionally incorporated of a “Sugoroku” board game featuring quizzes and interactive discussions regarding medicines. Participants aged ≥18 years who regularly used medicines took part in the program, which was conducted face-to-face under the guidance of a facilitator. Data were collected through questionnaire surveys administered before (baseline), immediately after (post), and one month following the program (follow-up). The surveys assessed the effectiveness of the program using the semantic differential method, along with changes in the participants’ awareness and behaviors related to medicine use. Eleven participants completed all the questionnaires, received high ratings across the ARCS model components. Notably, the number of participants who attempted to record their health information in a medication notebook increased from four at baseline to eight immediately after the program, with four participants continuing this behavior one month later. These findings suggest that the learning program contributed to positive changes in the awareness and behaviors of the participants regarding responsible medicine use.

BackgroundFamily genetic testing facilitates early cancer diagnosis and prevention for relatives of individuals carrying Breast Cancer Susceptibility Genes 1 and 2 (BRCA1/2) pathogenic variants. This study evaluates perceptions of family genetic testing among Korean cancer patients and the general public, providing foundational data to guide strategies for implementation.MethodsA total of 186 participants (86 patients with cancer and 100 public controls) were educated on hereditary breast and ovarian cancer and family genetic testing via a presentation video accessed through QR codes or smartphone links. Afterward, participants completed a 10-question knowledge test, and perceptions were assessed using a questionnaire featuring a hypothetical BRCA1/2 patient scenario to explore general characteristics and attitudes toward family genetic testing.ResultsPost-education, knowledge scores were significantly higher in the patient group than in the controls (median: 10 vs. 9, P = 0.040). Overall, 54.6% of participants shared their genetic test results, with patients sharing more frequently than controls (59.6% vs. 50.0%, P < 0.001). First-degree relatives were the most common recipients (patients: 81.0%, controls: 76.9%, P = 0.163). Results were shared to inform relatives of cancer risks and encourage prevention, while weak familial relationships were cited as barriers. Face-to-face communication was the preferred sharing method. Most participants (75.8%) believed genetic results should be shared with children once they turn 19.ConclusionPromoting family genetic testing requires effective communication and dissemination of accurate information. Developing systematic genetic counseling programs is essential to achieving these objectives.

This note introduces HaploThread, a user-friendly desktop software with a graphical user interface designed for constructing and visualizing haplotype networks. Developed in C++ using the Qt library, HaploThread integrates network visualization with multiple multithreaded haplotype construction algorithms-including McAN and fastHaN (which incorporates MSN, MJN, and TCS)-through a modular plugin architecture. It provides an intuitive workflow for building and visualizing haplotype networks from large-scale datasets, while also supporting functional extensions via plugins to facilitate the analysis of genetic variation and evolutionary relationships. HaploThread is released as open-source software under the GNU General Public License. Its source code and precompiled executables for Windows and macOS are freely available at https://ngdc.cncb.ac.cn/biocode/tool/BT007948 and https://github.com/git-xubo/HaploThread.

The institutionalization of public health in Spain

Background: Cardiopulmonary Resuscitation (CPR) is a critical emergency intervention that significantly improves survival during sudden cardiac arrest. In rural regions of India, including Garhwal, lack of CPR awareness among the general public remains a major concern. Objective: To assess the knowledge and impact of CPR training conducted in community pharmacy settings among the general public in Pokhra Block, Pauri District, Uttarakhand. Methodology: A prospective interventional pre-test and post-test study was conducted among 60 participants over a period of two months, including one month of CPR training. Knowledge and confidence levels were assessed using structured questionnaires before and after the intervention. Results: Post-training analysis showed a statistically significant improvement in CPR knowledge and confidence levels among participants. Conclusion: Community pharmacy-based CPR training is an effective strategy to enhance emergency preparedness among the general public in rural Uttarakhand.

The detection of gravitational waves from binary black hole and neutron star mergers by ground-based interferometers, as well as the evidence for a gravitational wave background from pulsar timing array experiments, has marked a new era in astrophysics and cosmology. These experiments also have great potential for discovering new physics through gravitational wave detection. One of the most motivated sources of gravitational waves that can be realized only within a beyond-the-Standard-Model framework is first-order phase transitions. In this work we release PT2GWFinder , a Mathematica package designed to compute phase transition parameters and the gravitational wave power spectrum for an arbitrary scalar theory exhibiting a first-order phase transition, in scenarios where a single scalar acquires a vacuum expectation value. PT2GWFinder performs the phase tracing, computes the bounce profile and action using FindBounce , calculates the relevant temperatures and phase transition parameters, and finally evaluates the gravitational wave spectrum. Additionally, it offers a user-friendly interface with DRalgo , which enables the computation of the dimensionally reduced effective potential in the high-temperature regime. This work includes a user manual and two models that demonstrate the capability and performance of PT2GWFinder . As a supplement, for one of these models we obtain the bounce solution and action analytically in the thin-wall approximation and demonstrate excellent agreement with the numerical approach. Program title : PT2GWFinder CPC Library link to program files: https://doi.org/10.17632/s69t25dw2j.1 Developer’s repository link : https://github.com/finshky/PT2GW Licensing provisions : GNU General Public License 3 Programming language : Mathematica Nature of problem : Search and characterization of cosmological first-order phase transitions, computation of the generated gravitational wave spectra. Solution method : Construction of the Euclidean bounce action function, using FindBounce , and application of integral criteria to determine the transition temperatures. Restrictions : Mathematica version 13 or above, applicable to single-field models.

Accurate annotation of coding sequences and translational features within transcript models is essential for interpreting assembled transcriptomes and their functional potential. Existing open reading frame (ORF) prediction tools typically operate on transcript FASTA files and do not reintegrate coding sequence (CDS) information back into transcript models, limiting their utility in long-read sequencing workflows where GTF/GFF annotations are the primary output. We present ORFannotate, a lightweight, GTF-native Python command-line tool that predicts ORFs from transcript annotations and reinserts precise, exon-aware CDS and UTR features into the original GTF/GFF file. In addition, ORFannotate provides biologically informative translational context by annotating Kozak sequence strength, detecting non-overlapping upstream ORFs (uORFs) with coding probabilities, characterising 5' and 3' untranslated regions (UTRs), and predicting nonsense-mediated decay (NMD) susceptibility. All annotations are consolidated in a transcript-level summary to support downstream analysis. By generating GTF files with accurate CDS annotations, ORFannotate facilitates reproducible analysis of both long- and short-read transcriptomes and integrates seamlessly with visualization tools, genome browsers, and comparative transcript analysis workflows. ORFannotate is fast, scalable and provides a practical solution for transcriptome annotation beyond coding potential prediction alone. ORFannotate is implemented in Python and freely available under the GNU General Public License v3 (GPL-3.0) at: https://github.com/egustavsson/ORFannotate (DOI: https://doi.org/10.5281/zenodo.16812866).

Incomplete gene models negatively impact single-cell gene expression quantification. This is particularly true in non-model species where often gene 3' ends are inaccurately annotated, while most scRNA-seq methods only capture the 3' transcript region. This results in many genes being incorrectly quantified or not detected. GeneExt leverages scRNA-seq data to refine gene annotations. We exemplify GeneExt usage and its impact on the gene expression quantification of eight non-model organism single-cell atlases. By extending and homogenizing gene annotations, our tool will help improve biological interpretation and cross-species comparisons of cell type expression atlases. GeneExt is available at https://github.com/sebepedroslab/GeneExt (DOI: https://doi.org/10.5281/zenodo.18712940) under a GNU General Public license, together with test data and usage instructions.

Introduction: In Switzerland, approximately CHF 25.9 Mrd. (3.2% of its GDP) were spent on research, development and innovation in 2023 (Körsgen et al., 2025), which is an increase of 1.3 Mrd. and a decrease of 0.2% in GDP relation compared to 2021 (Körsgen et al., 2023). The general publication of the FSO does not differentiate research and innovation in sports as a separate category. Due to the low ranking of Switzerland in an international comparison (Kempf; et al., 2021), a first analysis of publicly funded research and innovation projects in 2021 has revealed quantitative and qualitative insights. The analysis revealed a total funding of CHF 5 Mio by the Swiss National Science Foundation (SNSF) and 1.5 Mio by Innosuisse, were most funding was received by sports medicine project (Kempf et al., 2025). In order to provide a longitudinal comparison, this analysis was now replicated for the year 2023. Methods: The semantic keyword-based search from the analysis of data from 2022 was replicated from the previous analysis (Kempf et al., 2025), searching the project database of SNSF, and ARAMIS, a database containing all research and innovation projects funded or executed by the Swiss federal government. In addition, further variables were analyzed to provide a broader view on the status quo on publicly funded research and innovation in Swiss sport. Results: While the analysis is still in progress, currently available data does not show largely different results between the years 2022 and 2023. Already analyzed projects confirm the results of 2022 with a large part of research and innovation funding received by sports medicine projects. Regarding institutions involved, multiple research and innovation facilities were identified, which had not been present in the data of 2022. Non-profit sports organizations like clubs or (inter)national federations are not part of them. Discussion/Conclusion: While the final results are not available as of now, it can be stated that frequent and objective monitoring of research and innovation data in sports is helpful for policy makers in public and private sports. The results of this investigation allow a longitudinal comparison over multiple years. With recently introduced measures on innovation by Swiss Olympic and the Federal Office of Sport FOSPO, the upcoming changes to the federation support model by Swiss Olympic and the upcoming measures triggered by the federal strategy “Sports- and Physical Activity Promotion 2040”, the results from this analysis can help to identify potential positive and negative impacts on research and innovation in Swiss sports, as well as it’s representation in the general national innovation system of Switzerland. In addition, further investigations on privately funded research and innovation projects in sports are necessary to provide a holistic perspective. References Kempf, H., Gübeli, C., Ganné-Chedeville, C., &amp; Emmenegger, S. (2025). The role of sport in research and development in Switzerland - An overview. Current Issues in Sport Science, 10(2), 006. https://doi.org/https://doi.org/10.36950/2025.2ciss006 Kempf;, H., Weber;, A., Zurmühle;, C., Bosshard;, B., Mrkonjic;, M., Weber;, A., Pillet;, F., &amp; Sutter, S. (2021). Leistungssport Schweiz – Momentaufnahme SPLISS-CH 2019. B. f. S. BASPO. Körsgen, A., Chardon, S. P., &amp; Sollberger, P. (2023). Forschung und Entwicklung in der Schweiz 2021. B. f. Statistik. https://dam-api.bfs.admin.ch/hub/api/dam/assets/24905944/master Körsgen, A., Plaza Chardon, S., &amp; Sollberger, P. (2025). Forschung und Entwicklung in der Schweiz 2023. Bundesamt für Statistik. https://doi.org/https://doi.org/10.71668/qdr9-sc23

Retinitis pigmentosa (RP) is a progressive and sight-threatening condition with few treatment options. Patients with RP experience variable progressive vision deterioration that can severely affect functioning. Currently, there is no published literature to indicate the preferences that patients with RP place on potential treatments. This study compared patient preferences to those of a sample from the general public. An online, discrete choice experiment (DCE) elicited preferences of a United States (US) sample for maintaining visual function-related activities of daily living (VF-ADL). The data were stratified by sociodemographic and clinical characteristics. Participants completed best-/worst-choice tasks using predefined attributes and varying levels, including time to deterioration in mobility and ability to read associated with worsening visual function. A total of 150 patients with RP and 301 public participants completed the DCE. By the age of 32-40 years, VF-ADL deteriorated substantially in patients versus public (81% vs 13% reported lower combined mobility/ability to read). Patients with RP placed the greatest value on ability to get around (relative attribute importance [RAI] 51%; standard error [SE] 3.4; 95% confidence interval [CI] 45-58), whereas the public sample preferred ability to get information (RAI 48%; SE 4.7; 95% CI 39-57). Patients with RP placed less weight on extending life expectancy compared with public participants. Patients with RP demonstrated preferences in VF-ADL and life expectancy associated with deterioration in visual function and potential treatment benefits that differ from those of the public. Considering patient preferences is critical to understand disease burden and estimate the value of therapeutic interventions.

The plasma membrane and accompanying cortex serve as one of the major hubs of the signal transduction and cytoskeletal activities that collectively regulate numerous cell physiological processes such as migration, polarity, macropinocytosis, phagocytosis, cytokinesis, etc. Yet, dynamically tracking membrane-cortex associated protein or lipid kinetics over time from live-cell image series remains a challenging task, primarily due to the difficulty of accurately extracting and aligning the cell boundary between consecutive frames, as the cell continuously deforms and moves. Here, we present Membrane Kymograph Generator , a cross-platform software that accepts multichannel time-lapse live-cell fluorescent imaging datasets as input and automates the cumbersome, heuristic process of boundary tracking, inter-frame alignment, and intensity sampling along the boundary. The software implements a rotational offset minimization algorithm that circularly aligns boundaries across consecutive frames by exhaustively searching for the optimal angular shift that minimizes point-to-point distances, while handling variations in boundary point counts due to cell shape changes. The software outputs kymographs that represent spatiotemporal dynamics of different membrane-associated proteins or biosensors, allows users to fine-tune visualization parameters through an interactive interface, and provides built-in correlation analysis tools for multi-channel datasets. Furthermore, the software allows advanced programmatic usage for batch processing and further analysis via a native API. Our validation tests demonstrated that the Membrane Kymograph Generator can be used to accurately track, visualize, and quantitate the spatial kinetics of a wide array of membrane proteins and lipid biosensors over extended time periods, in a variety of cell types, including Dictyostelium amoeba, human neutrophils, mouse macrophages, and different mammalian cancer cells. The GUI-based software is user-friendly, does not require any technical expertise from users, and significantly reduces the manual effort and time required for kymograph generation and downstream analysis, while ensuring high accuracy and reproducibility. Membrane Kymograph Generator is a free and open-source software, licensed under GNU General Public License 3.0 or later. This software is cross-platform: It can be graphically installed on both x86-64 and AArch64/ARM64 computers, running either Windows, macOS, or any standard Linux distribution. The software is distributed as single installer files (and portable executables) targeting specific hardware architectures and operating systems, and hence, it can be installed natively without any dependency resolution. The source code, detailed documentation, specific installers, portable binaries, and test data are freely available at https://github.com/tatsatb/membrane-kymograph-generator . Additionally, since the software is written in Python, it can also be installed inside any Python environment using PIP package manager (package ID: https://pypi.org/project/membrane-kymograph ) and can also be interacted via a built-in Python API.

Hydration thermodynamic quantities are essential for understanding protein function from a free-energy perspective. The grid inhomogeneous solvation theory (GIST) enables the computation of spatial distributions of hydration energy, ΔEW(r), and hydration entropy, ΔSW(r), using molecular dynamics (MD) simulations, from which the distribution of the hydration free energy, ΔGW(r), is obtained as ΔGW(r) = ΔEW(r) - TΔSW(r), where T is the absolute temperature. However, GIST is computationally demanding, requiring tens of hours to compute these distributions. To overcome this bottleneck, we developed a set of deep learning models capable of predicting ΔEW(r), TΔSW(r), and ΔGW(r). Our deep learning models completed these predictions within tens of seconds using a single graphics processing unit. The resulting distributions achieved coefficient of determination values of 0.76-0.84 for ΔGW(r) when compared to GIST results, and lower values were obtained for ΔEW(r) and TΔSW(r). As a practical application, we examined the free energy change required for a water molecule to move from the bulk region to the ligand-binding site, ΔGW,replace, using both our deep learning model and GIST. A high correlation coefficient of 0.78 was observed between the predictions of our model and GIST, confirming its reliability. Furthermore, the results for a representative protein were consistent with experimental data of the corresponding protein-ligand complex: Water molecules with low ΔGW,replace values located near crystallographic waters, suggesting retention upon ligand binding, whereas those with unfavorable values overlapped with the ligand, indicating displacement upon the ligand binding. These findings demonstrate that our deep learning models provide an efficient and accurate alternative to GIST for predicting hydration thermodynamics and enable the consideration of protein conformational fluctuations, which is difficult to achieve with conventional GIST. The program called "Deep GIST" is available under the GNU General Public License from https://github.com/YoshidomeGroup-Hydration/Deep-GIST.

Network theory has proven invaluable in unraveling complex protein interactions. Previous studies have employed statistical methods rooted in network theory, including the Gaussian graphical model, to infer networks among proteins, identifying hub proteins based on key structural properties of networks such as degree centrality. However, there has been limited research examining a prognostic role of hub proteins on outcomes, while adjusting for clinical covariates in the context of high-dimensional data. To address this gap, we propose a network-guided penalized regression method. First, we construct a network using the Gaussian graphical model to identify hub proteins. Next, we preserve these identified hub proteins along with clinically relevant factors, while applying adaptive Lasso to non-hub proteins for variable selection. Our network-guided estimators are shown to have variable selection consistency and asymptotic normality. Simulation results suggest that our method produces better results compared to existing methods and demonstrates promise for advancing biomarker identification in proteomics research. Lastly, we apply our method to the Clinical Proteomic Tumor Analysis Consortium (CPTAC) data and identified hub proteins that may serve as prognostic biomarkers for various diseases, including rare genetic disorders and immune checkpoint for cancer immunotherapy. R package is freely available on CRAN repository (https://CRAN.R-project.org/package=NetGreg) and published under General Public License version 3.