Search

Abstract Resistance to pod shattering is a key domestication-related trait selected for seed production in many crops. Here, we show that the transition from shattering in wild soybeans to shattering resistance in cultivated soybeans resulted from selection of mutations within the coding sequences of two nearby genes - Sh1 and Pdh1. Sh1 encodes a C2H2-like zinc finger transcription factor that promotes shattering by repressing SHAT1-5 expression, thereby reducing the secondary wall thickness of fiber cap cells in the abscission layers of pod sutures, while Pdh1 encodes a dirigent protein that orchestrates asymmetric lignin distribution in inner sclerenchyma, creating torsion in pod walls that facilitates shattering. Integration analyses of quantitative trait locus mapping, genome-wide association studies, and allele distribution in representative soybean germplasm suggest that these two genes are primary modulators underlying this domestication trait. Our study thus provides comprehensive understanding regarding the genetic, molecular, and cellular bases of shattering resistance in soybeans.

As advanced silicon manufacturing processes slow down in scaling, three-dimensional stacking of chips has become a future trend in microelectronic industry. High-density bonding technology is one of the core technologies used to achieve the stacking. In this work, we explored a new bonding technology that is possible to manufacture future high-density interconnects. We fabricated a series of sandwich structure solder joints with different thicknesses using SnBiIn-based nanoparticles under a low bonding temperature of 100 °C. The intermetallic compound (IMC) at the interface is proven to be η′-Cu6(Sn, In)5 nanocrystals with random grain orientations through X-ray diffraction (XRD), high-resolution transmission electron microscope (HRTEM), and atomic probe tomography (APT). The hardness and elastic modulus of η′-Cu6(Sn, In)5 were measured to be 4.093 GPa and 79.586 GPa by micropillar compression tests, respectively, decreasing by 35% and 29.1% compared to Cu6Sn5. Therefore, doping In atoms reduces the brittleness of IMCs, providing a new approach for mitigating the mechanical failures of full IMC bump. First-principles calculation also verified the change in mechanical properties. Furthermore, a full-Cux(Sn, In)y IMCs solder joint and an array of 3 μm solder joints were fabricated experimentally using the bonding technology. This technology successfully demonstrates a pioneer interconnection bonding methodology for future ultra-high-density chip stacking using low melting point solder nanoparticles.

Abstract Purpose This study aimed to analyse the correlation between blood glucose control and the severity of COVID-19 infection in patients with diabetes. Methods Clinical and imaging data of a total of 146 patients with diabetes combined with COVID-19 who visited our hospital between December 2022 and January 2023 were retrospectively collected. The patients were divided into the ‘good blood glucose control’ group and the ‘poor blood glucose control’ group based on an assessment of their blood glucose control. The clinical data, computed tomography (CT) appearance and score and the severity of COVID-19 infection of the two groups were compared, with the severity of COVID-19 infection being the dependent variable to analyse other influencing factors. Results The group with poor blood glucose control showed a higher lobar involvement degree and total CT severity score (CTSS) than the group with good blood glucose control (13.30 ± 5.25 vs. 10.38 ± 4.84, p 0.05). The area under the curve (AUC) of CTSS prediction of critical COVID-19 was 0.895 with sensitivity of 79.3% and specificity of 88.1% when the threshold value is 12. Conclusion Blood glucose control is significantly correlated with the CTSS; the higher the blood glucose is, the more severe the lung manifestation. The CTSS can also be used to evaluate and predict the clinical severity of COVID-19.

The subjective encounter with architecture encompasses the particular presentation of architectural elements or the overall structure to individuals, taking into account their perception, cognition, and thought processes. The purpose of this research is to investigate the intricate relationship between visual attention directed towards historical architecture and the subjective experiences engendered during the process of observation. Utilizing eye-tracking technology within the realm of virtual reality, this study delves into the observation patterns exhibited by individuals when confronted with historical architecture, specifically focusing on the traditional courtyard residences found in Beijing’s hutongs. The panoramic images, meticulously modeled and rendered, are divided into six distinct areas of interest: base, ground, window&doors, walls, roof, and eaves. The eye-tracking data of 81 participants, who engaged with 10 virtual scenes through the employment of VR headsets, along with their responses to architectural style questionnaires, were systematically gathered. Through comprehensive analysis, encompassing the examination of total fixation duration (TFD), fixation count (FC), first fixation duration (FFD), and time to first fixation (TFF) across the areas of interest (AOIs), in addition to deviations in scores among different architectural areas and styles, notable insights emerged. Results indicate that both professionals and non-professionals allocate heightened attention, as evidenced by TFD, FC, TFF, and FFD, to elements that exhibit greater score deviations in the questionnaires as opposed to those with smaller discrepancies. Moreover, the TFD and FC metrics pertaining to the Windows&doors AOI collectively constitute 40.20% and 40.71%, respectively. Undoubtedly, these figures signify the preeminent focal points within all AOIs. This underscores the pivotal role played by doors and windows in shaping individuals’ judgments pertaining to historical architectural styles. This research provides designers with valuable understanding of the cognitive patterns exhibited by individuals in engaging with aspects of history, which enable them to identify and preserve key elements associated with historical styles. Additionally, it establishes a fundamental cornerstone for future quantitative research endeavors centered around the preservation of historical architectural styles.

Abstract Background There is an urgent clinical need for developing novel immunoprophylaxis and immunotherapy strategies against Staphylococcus aureus (S. aureus). In our previous work, immunization with a tetra-branched multiple antigenic peptide, named MAP2-3 that mimics lipoteichoic acid, a cell wall component of S. aureus, successfully induced a humoral immune response and protected BALB/c mice against S. aureus systemic infection. In this study, we further investigated whether vaccination with MAP2-3 can elicit immunologic memory. Methods BALB/c mice were immunized with MAP2-3 five times. After one month of the last vaccination, mice were challenged with heat-killed S. aureus via intraperitoneal injection. After a 7-day inoculation, the percentage of plasma cells, memory B cells, effector memory T cells, and follicular helper T cells were detected by flow cytometry. The levels of IL-6, IL-21, IL-2, and IFN-γ were measured by real-time PCR and ELISA. Flow cytometry results were compared by using one-way ANOVA or Mann-Whitney test, real-time PCR results were compared by using one-way ANOVA, and ELISA results were compared by using one-way ANOVA or student’s t-test. Results The percentage of plasma cells and memory B cells in the spleen and bone marrow from the MAP2-3 immunized mice was significantly higher than that from the control mice. The percentage of effector memory T cells in spleens and lymphoid nodes as well as follicular helper T cells in spleens from the MAP2-3 immunized mice were also higher. Moreover, the levels of IL-6 and IL-21, two critical cytokines for the development of memory B cells, were significantly higher in the isolated splenocytes from immunized mice after lipoteichoic acid stimulation. Conclusions Immunization with MAP2-3 can efficiently induce memory B cells and memory T cells.

Abstract High-grade serous ovarian carcinoma (HGSOC) is genetically unstable and characterised by the presence of subclones with distinct genotypes. Intratumoural heterogeneity is linked to recurrence, chemotherapy resistance, and poor prognosis. Here, we use spatial transcriptomics to identify HGSOC subclones and study their association with infiltrating cell populations. Visium spatial transcriptomics reveals multiple tumour subclones with different copy number alterations present within individual tumour sections. These subclones differentially express various ligands and receptors and are predicted to differentially associate with different stromal and immune cell populations. In one sample, CosMx single molecule imaging reveals subclones differentially associating with immune cell populations, fibroblasts, and endothelial cells. Cell-to-cell communication analysis identifies subclone-specific signalling to stromal and immune cells and multiple subclone-specific autocrine loops. Our study highlights the high degree of subclonal heterogeneity in HGSOC and suggests that subclone-specific ligand and receptor expression patterns likely modulate how HGSOC cells interact with their local microenvironment.

Zero-knowledge proof (ZKP) is a cryptographic primitive that enables a prover to convince a verifier that a statement is true, without revealing any other information beyond the correctness of the statement itself. Due to its powerful capabilities, its most practical type, called zero-knowledge Succinct Non-interactive ARgument of Knowledge (zkSNARK), has been widely deployed in various privacypreserving applications such as cryptocurrencies and verifiable computation. Although state-of-the-art zkSNARKs are highly efficient for the verifier, the computational overhead for the prover is still orders of magnitude too high to warrant use in many applications. This overhead arises from several time-consuming operations, including large-scale matrix-vector multiplication (MUL), number-theoretic transform (NTT), and especially the multi-scalar multiplication (MSM) which constitutes the largest proportion. Therefore, further efficiency improvements are needed. In this paper, we focus on comprehensive optimization of running time and storage space required by the MSM algorithm on GPUs. Specifically, we propose a novel, modular and adaptive parameter configuration technique—elastic MSM to enable us to adjust the scale of MSM according to our own wishes by performing a corresponding amount of preprocessing. This technique enables us to fully unleash the potential of various efficient parallel MSM algorithms. We have implemented and tested elastic MSM over three prevailing parallel Pippenger algorithms on GPUs. Across various preprocessing space limitations (across various MSM scales), our constructions achieve up to about 1.90×, 1.08× and 1.36× (2.58×, 1.39× and 1.91×) speedup versus three state-of-the-art parallel Pippenger algorithms on GPUs, respectively. From another perspective, elastic MSM could also be regarded as a preprocessing technique over the well-known Pippenger algorithm, which is modular and could be used to accelerate almost all the most advanced parallel Pippenger algorithms on GPUs. Meanwhile, elastic MSM provides an adaptive trade-off between the running time and the extra storage space needed by parallel Pippenger algorithms on GPUs. This is the first preprocessing technique to retain the improved MSM computation brought by preprocessing under varying storage space limitations. Specifically, across various preprocessing space limitations (across various MSM scales), our constructions achieve up to about 192× and 223× (159× and 174×) speedup versus two state-ofthe- art preprocessing parallel Pippenger algorithms on GPUs, respectively.

BackgroundAn escalating body of clinical trials and observational studies hints at a plausible link between gut flora and postpartum depression (PPD). The definitive causal dynamics between these two entities remain shrouded in ambiguity. Therefore, in this study, we employed the two-sample Mendelian randomization approach to ascertain the causal link between gut microbiota and PPD.MethodsSummary-level GWAS data related to the human gut microbiota were obtained from the international consortium MiBioGen and the Dutch Microbiome Project (species). For PPD, GWAS data were derived from the FinnGen biobank, consisting 57,604 cases and 596,601 controls. The inverse variance weighted method (IVW) as the cornerstone of our analytical approach. Subsequent to this, a comprehensive suite of tests for pleiotropy and heterogeneity were conducted to ensure the reliability and robustness of our findings.ResultsWe identified 12 bacterial taxa associated with the risk of PPD. Veillonellaceae, Ruminococcaceae UCG 011, Bifidobacterium adolescentis, Paraprevotella clara, Clostridium leptum, Eubacterium siraeum, Coprococcus catus exhibited an inversely associated with the risk of PPD. Alphaproteobacteria, Roseburia, FamilyXIIIAD3011group, Alistipes onderdonkii, Bilophila wadsworthia showed a positive correlation with the risk of PPD.LimitationsThe GWAS data derived from the MiBioGen consortium, DMP, and FinnGen consortium, may introduce selection bias. Moreover, the data primarily originates from European populations, hence extrapolating these results to diverse populations should be approached with caution. The etiological factors behind PPD remain enigmatic, alluding to the existence of potential undisclosed confounders.ConclusionBased on this MR analysis, we found a causal relationship between certain gut microbial communities and PPD. Future clinical studies can further explore the treatment of PPD through the combined use of microorganisms. This not only offers insights into the pathogenesis of PPD but also lays the foundation for utilizing gut microbiota as biotherapeutics in treating neurological disorders.

Introduction First language care is critical for older immigrant adults with limited English proficiency, especially in long-term care settings where most residents require staff assistance and experience complex chronic conditions, resulting in multiple communication interactions where language poses a barrier. Although there are a myriad of cultural-language translation apps and devices available, there is a gap in both research and practice on the acceptability and feasibility of these digital resources within the context of long-term care and community settings for older immigrant adults, from a cultural relevance and digital health equity perspective. Our paper outlines a scoping review protocol to examine the state of the literature on the extent to which cultural-language translation apps are used in long-term care settings and community-based elder care. We will also examine the extent to which such apps bridge or further gaps in equitable, accessible and acceptable care for older immigrant adults with limited English language proficiency.Methods and analysis This scoping review protocol will employ an adapted five-stage framework outlined by Arksey and O’Malley guided by enhancements recommended by Levac et al and Colquhoun et al. Using the Joanna Briggs Institute’s population, concept and context framework, we defined the scope of the scoping review by identifying the target population, concepts for investigation and the context within which the research is situated. We will conduct a search of the literature from 2005 to 2024 using five bibliographic databases from health sciences (Healthstar OVID, MEDLINE OVID and Cumulative Index to Nursing and Allied Health Literature (CINAHL) EBSCO), engineering (Engineering Village Elsevier) and a cross-disciplinary database (Web of Science Clarivate). The research team will adopt a critical, equity-focused approach for the scoping review by integrating Richardson et al’s framework for Digital Health Equity into our analysis of the findings. This will ensure that health and social equity perspectives are integrated within our methodology and analytical lens. Our analysis will specifically examine selected studies for their engagement with health equity and their ability to address issues such as ageism, ableism and the digital divide within geriatric care.Ethics and dissemination Ethics approval is not required for this scoping review as it involves secondary analysis of published works and no primary data collection involving human subjects. Findings of the review will be shared with community partners and disseminated through publications, conferences and peer-reviewed publications.

The role of the gut microbiota in host physiology has been previously elucidated for some marine organisms, but little information is available on their metabolic activity involved in transformation of environmental pollutants. This study assessed the metabolic profiles of the gut microbial cultures from grouper (Epinephelus coioides), green mussel (Perna viridis) and giant tiger prawn (Penaeus monodon) and investigated their transformation mechanisms to typical plastic additives. Community-level physiological profiling analysis confirmed the utilization profiles of the microbial cultures including carbon sources of carbohydrates, amines, carboxylic acids, phenolic compounds, polymers and amino acids, and the plastic additives of organophosphate flame retardants, tetrabromobisphenol A derivates and bisphenols. Using in vitro incubation, triphenyl phosphate (TPHP) was found to be rapidly metabolized into diphenyl phosphate by the gut microbiota as a representative ester-containing plastic additive, whereas the transformation of BPA (a representative phenol) was relatively slower. Interestingly, all three kinds of microbial cultures efficiently transformed the hepatic metabolite of BPA (BPA-G) back to BPA, thereby increasing its bioavailability in the body. The specific enzyme analysis confirmed the ability of the gut microbiota to perform the metabolic reactions. The results of 16S rRNA sequencing and network analysis revealed that the genera Escherichia-Shigella, Citrobacter, and Anaerospora were functional microbes, and their collaboration with fermentative microbes played pivotal roles in the transformation of the plastic additives. The structure-specific transformations by the gut microbiota and their distinct bioavailability deserve more attention in the future.

Bootstrapping is a critical technique in constructing fully homomorphic encryption (FHE), which serves to refresh the noise in FHE ciphertexts, enabling an arbitrary number of homomorphic operations. Among published results, the TFHE-rs library [Zam22] offers the fastest bootstrapping implementation on CPU platforms by taking advantage of AVX-512 instructions. In this paper, we improve the efficiency of the bootstrapping algorithm based on the NTRU problem. First, we introduce the approximate gadget decomposition method tailored for NTRU ciphertext, reducing the number of NTT operations required for external products. Second, by integrating the approximate decomposition and key unrolling techniques, we improve the performance of CMux-based blind rotation. Third, for the automorphism-based blind rotation method, we present a hybrid window size technique that reduces the number of automorphisms by 34% compared to recent work [XZD+23](in Crypto23). Subsequently, we implement the proposed bootstrapping algorithm on the CPU platform with AVX instructions. Experimental results demonstrate that our method only takes 3.8ms, which achieves a 1.8× speedup compared to the TFHE-rs library. Finally, we propose an efficient FPGA accelerator based on the CMux method, which not only achieves the best performance but also exhibits high throughput advantages. Our accelerator can improve performance by 2x compared to state-of-the-art FPGA implementations (e.g., FPT).

Abstract Objective To clarify whether the 3D printing model has auxiliary functions in toto extraction of donor tooth in autotransplantation cases. Methods Two hundred and sixty patients who would have operation of ATT were divided into two groups. In group 1, determination of the tooth extraction in toto was predicted only according to the clinical and imaging examination. In group 2, the prediction was performed according to the clinical and imaging examination as well as the 3D model of donor tooth pre-extraction. A prespctive clinical study was designed on intra-group comparison between the predicted and actual donor teeth situation when extraction in cases of ATT. The consistent rate for the predicted results and the actual results were compared with the two groups. Results A remarkable difference was observed between the predicted results and the actual results of tooth positions and root numbers in group without model (p

Abstract Background Adoptive cell therapy has achieved great success in treating hematological malignancies. However, the production of chimeric antigen receptor T (CAR-T) cell therapy still faces various difficulties. Natural killer (NK)-92 is a continuously expandable cell line and provides a promising alternative for patient’s own immune cells. Methods We established CAR-NK cells by co-expressing natural killer group 2 member D (NKG2D) and IL-21, and evaluated the efficacy of NKG2D-IL-21 CAR-NK cells in treating lung cancer in vitro and in vivo. Results Our data suggested that the expression of IL-21 effectively increased the cytotoxicity of NKG2D CAR-NK cells against lung cancer cells in a dose-dependent manner and suppressed tumor growth in vitro and in vivo. In addition, the proliferation of NKG2D-IL-21 CAR-NK cells were enhanced while the apoptosis and exhaustion of these cells were suppressed. Mechanistically, IL-21-mediated NKG2D CAR-NK cells function by activating AKT signaling pathway. Conclusion Our findings provide a novel option for treating lung cancer using NKG2D-IL-21 CAR-NK cell therapy.

Abstract Objective To evaluate the effectiveness, safety, and convenience of in‐class transition (iCT) from intravenous bortezomib‐based induction to ixazomib‐based oral regimens. Methods This retrospective real‐world study was conducted in 16 Chinese hospitals between October 2017 and April 2023 and analyzed newly diagnosed (NDMM) and first‐line relapsed multiple myeloma (FRMM) patients who attained at least a partial response from bortezomib‐based induction therapy, followed by an ixazomib‐based oral regimen for 2 year or until disease progression or intolerable toxicity. Results The study enrolled 199 patients, median age: 63 years old, male 55.4%, 53% as high risk (HR), and 47% as standard risk. Cytogenetic risk stratification by metaphase fluorescence in situ hybridization (M‐FISH), based on the Mayo Clinic risk stratification system. The median duration of total PI therapy was 11 months, with ixazomib‐based treatment spanning 6 months. At the 20‐month median follow‐up, 53% of patients remained on therapy. The 24‐month PFS rate was 84.3% from the initiation of bortezomib‐based induction and 83.4% from the start of ixazomib‐based treatment. Overall response rate (ORR) was 100% post‐bortezomib induction and 90% following 6 cycles of the ixazomib‐based regimen. Based on the Sankey diagrams, 89.51% of patients maintained or improved their disease response after 2 cycles of iCT, 6 cycles (90.14%), and 12 cycles (80%). The HR level of Mayo was found to be a significant independent factor in a worse remission (hazard ratio (HR) 2.55; p = 0.033). Ixazomib's safety profile aligned with previous clinical trial data, with 49% of patients experiencing at least one AE of any grade. The most common AEs included peripheral neuropathy, nausea and vomiting, diarrhea, thrombocytopenia, and granulocytopenia. Conclusion In the real‐world Chinese MM population, NDMM and FRMM patients responded favorably to PI‐based continuous therapy, demonstrating substantial response rates. The ixazomib‐based iCT allows for sustained PI‐based treatment, offering promising efficacy and tolerable AEs.

CRISPR-based screens have discovered novel functional genes involving in diverse tumor biology and elucidated the mechanisms of the cancer pathological states. Recently, with its randomness and unbiasedness, CRISPR screens have been used to discover effector genes with previously unknown roles for AML. Those novel targets are related to AML survival resembled cellular pathways mediating epigenetics, synthetic lethality, transcriptional regulation, mitochondrial and energy metabolism. Other genes that are crucial for pharmaceutical targeting and drug resistance have also been identified. With the rapid development of novel strategies, such as barcodes and multiplexed mosaic CRISPR perturbation, more potential therapeutic targets and mechanism in AML will be discovered. In this review, we present an overview of recent progresses in the development of CRISPR-based screens for the mechanism and target identification in AML and discuss the challenges and possible solutions in this rapidly growing field.

Abstract This paper presents an enhanced whale optimization algorithm based on virtual force to optimize coverage and address the problem of uneven coverage during the deployment of underwater sensor networks. This method is guided by the Lévy flight and virtual force algorithms and adopts a differential mutation strategy based on random walks. This improves the fitness of the initial population and population richness of the algorithm. The effectiveness of this method was verified through simulation.

Abstract In clinical trials involving patients with HER2 (ERBB2 receptor tyrosine kinase 2) positive gastric cancer, the efficacy of the HER2-targeted drug lapatinib has proven to be disappointingly poor. Under the persistent pressure exerted by targeted drug therapy, a subset of tumor cells exhibit acquired drug resistance through the activation of novel survival signaling cascades, alongside the proliferation of tumor cells that previously harbored mutations conferring resistance to the drug. This study was undertaken with the aim of elucidating in comprehensive detail the intricate mechanisms behind adaptive resistance and identifying novel therapeutic targets that hold promise in the development of effective lapatinib-based therapies for the specific subset of patients afflicted with gastric cancer. We have successfully established a gastric cancer cell line with acquired lapatinib resistance, designated as HGC-27-LR cells. Utilizing comprehensive coding and noncoding transcriptome sequencing analysis, we have identified key factors that regulate lapatinib resistance in HGC-27 cells. We have compellingly validated that among all the lncRNAs identified in HGC-27-LR cells, a novel lncRNA (long noncoding RNA) named NONHSAT160169.1 was found to be most notably upregulated following exposure to lapatinib treatment. The upregulation of NONHSAT160169.1 significantly augmented the migratory, invasive, and stemness capabilities of HGC-27-LR cells. Furthermore, we have delved into the mechanism by which NONHSAT160169.1 regulates lapatinib resistance. The findings have revealed that NONHSAT160169.1, which is induced by the p-STAT3 (signal transducer and activator of transcription 3) nuclear transport pathway, functions as a decoy that competitively interacts with hsa-let-7c-3p and thereby abrogates the inhibitory effect of hsa-let-7c-3p on SOX2 (SRY-box transcription factor 2) expression. Hence, our study has unveiled the NONHSAT160169.1/hsa-let-7c-3p/SOX2 signaling pathway as a novel and pivotal axis for comprehending and surmounting lapatinib resistance in the treatment of HER2-positive gastric cancer.

Estuaries serve as crucial filters for land-based pollutants to the open sea, but there is a lack of information on the migration and fate of organophosphate flame retardants (OPFRs) within estuaries. This study focused on the Pearl River Estuary (PRE) by examining the co-occurrence of OPFRs and their metabolites and quantifying their transport fluxes using a mass balance model. The seawater concentrations of OPFRs and their metabolites exhibited significant seasonal variations (p

Pod dehiscence brings much loss for modern agricultural production, and multiple pod dehiscence components have been identified in many plant species. However, the pod dehiscence regulation factors in soybean are limited. In this study, we investigate the function of GmDIR26, a close homologues gene of pod dehiscence genes GmPdh1, PvPdh1, and CaPdh1, in the regulation of pod dehiscence in soybean. The secondary and tertiary structure analysis reveals that GmDIR26 protein has a similar structure with GmPdh1, PvPdh1, and CaPdh1 proteins. Synteny analysis of soybean and chickpea genomes shows that the genomic region surrounding GmDIR26 and CaPdh1 might be evolved from the same ancestor, and these two genes might have similar function. GmDIR26 shows an increased expression pattern during pod development and reaches a peak at beginning seed stage. Meanwhile, GmDIR26 exhibits high expression levels in dorsal suture and pod wall, but low expression pattern in ventral suture. In addition, GmDIR26 shows higher expression levels in pod dehiscence genotype than that in pod indehiscence accessions. Overexpression of GmDIR26 in soybean increases pod dehiscence in transgenic plants, of which the lignin layer in inner sclerenchyma pods is thicker and looser. The expression levels of several pod dehiscence genes are altered. Our study provides important information for further modification of pod dehiscence resistance soybean and characterization of soybean pod dehiscence regulation network.

ObjectivesTo assess the performance bias caused by sampling data into training and test sets in a mammography radiomics study.MethodsMammograms from 700 women were used to study upstaging of ductal carcinoma in situ. The dataset was repeatedly shuffled and split into training (n = 400) and test cases (n = 300) forty times. For each split, cross-validation was used for training, followed by an assessment of the test set. Logistic regression with regularization and support vector machine were used as the machine learning classifiers. For each split and classifier type, multiple models were created based on radiomics and/or clinical features.ResultsArea under the curve (AUC) performances varied considerably across the different data splits (e.g., radiomics regression model: train 0.58-0.70, test 0.59-0.73). Performances for regression models showed a tradeoff where better training led to worse testing and vice versa. Cross-validation over all cases reduced this variability, but required samples of 500+ cases to yield representative estimates of performance.ConclusionsIn medical imaging, clinical datasets are often limited to relatively small size. Models built from different training sets may not be representative of the whole dataset. Depending on the selected data split and model, performance bias could lead to inappropriate conclusions that might influence the clinical significance of the findings.Advances in knowledgePerformance bias can result from model testing when using limited datasets. Optimal strategies for test set selection should be developed to ensure study conclusions are appropriate.

Abstract Background RNA binding proteins (RBPs)—regulated gene expression play a vital role in various pathological processes, including the progression of cancer. However, the role of RBP in hepatocellular carcinoma (HCC) remains much unknown. In this study, we aimed to explore the contribution of RBP CCDC137 in HCC development. Methods We analyzed the altered expression level and clinical significance of CCDC137 in database and HCC specimens. In vitro cell assays and in vivo spontaneous mouse models were used to assess the function of CCDC137. Finally, the molecular mechanisms of how CCDC137 regulates gene expression and promotes HCC was explored. Results CCDC137 is aberrantly upregulated in HCC and correlates with poor clinical outcomes in HCC patients. CCDC137 markedly promoted HCC proliferation and progression in vitro and in vivo. Mechanistically, CCDC137 binds with FOXM1, JTV1, LASP1 and FLOT2 mRNAs, which was revealed by APOBEC1-mediated profiling, to increase their cytoplasmic localization and thus enhance their protein expressions. Upregulation of FOXM1, JTV1, LASP1 and FLOT2 subsequently synergistically activate AKT signaling and promote HCC. Interestingly, we found that CCDC137 binds with the microprocessor protein DGCR8 and DGCR8 has a novel non-canonical function in mRNA subcellular localization, which mediates the cytoplasmic distribution of mRNAs regulated by CCDC137. Conclusions Our results identify a critical proliferation-related role of CCDC137 and reveal a novel CCDC137/DGCR8/mRNA localization/AKT axis in HCC progression, which provide a potential target for HCC therapy.