Robust Aggregation for Federated Learning by Minimum γ-Divergence Estimation.

Federated learning is a framework for multiple devices or institutions, called local clients, to collaboratively train a global model without sharing their data. For federated learning with a central server, an aggregation algorithm integrates model information sent from local clients to update the parameters for a global model. Sample mean is the simplest and most commonly used aggregation method. However, it is not robust for data with outliers or under the Byzantine problem, where Byzantine clients send malicious messages to interfere with the learning process. Some robust aggregation methods were introduced in literature including marginal median, geometric median and trimmed-mean. In this article, we propose an alternative robust aggregation method, named γ-mean, which is the minimum divergence estimation based on a robust density power divergence. This γ-mean aggregation mitigates the influence of Byzantine clients by assigning fewer weights. This weighting scheme is data-driven and controlled by the γ value. Robustness from the viewpoint of the influence function is discussed and some numerical results are presented.

Microenvironment-induced PTEN loss by exosomal microRNA primes brain metastasis outgrowth.

The development of life-threatening cancer metastases at distant organs requires disseminated tumour cells' adaptation to, and co-evolution with, the drastically different microenvironments of metastatic sites. Cancer cells of common origin manifest distinct gene expression patterns after metastasizing to different organs. Clearly, the dynamic interaction between metastatic tumour cells and extrinsic signals at individual metastatic organ sites critically effects the subsequent metastatic outgrowth. Yet, it is unclear when and how disseminated tumour cells acquire the essential traits from the microenvironment of metastatic organs that prime their subsequent outgrowth. Here we show that both human and mouse tumour cells with normal expression of PTEN, an important tumour suppressor, lose PTEN expression after dissemination to the brain, but not to other organs. The PTEN level in PTEN-loss brain metastatic tumour cells is restored after leaving the brain microenvironment. This brain microenvironment-dependent, reversible PTEN messenger RNA and protein downregulation is epigenetically regulated by microRNAs from brain astrocytes. Mechanistically, astrocyte-derived exosomes mediate an intercellular transfer of PTEN-targeting microRNAs to metastatic tumour cells, while astrocyte-specific depletion of PTEN-targeting microRNAs or blockade of astrocyte exosome secretion rescues the PTEN loss and suppresses brain metastasis in vivo. Furthermore, this adaptive PTEN loss in brain metastatic tumour cells leads to an increased secretion of the chemokine CCL2, which recruits IBA1-expressing myeloid cells that reciprocally enhance the outgrowth of brain metastatic tumour cells via enhanced proliferation and reduced apoptosis. Our findings demonstrate a remarkable plasticity of PTEN expression in metastatic tumour cells in response to different organ microenvironments, underpinning an essential role of co-evolution between the metastatic cells and their microenvironment during the adaptive metastatic outgrowth. Our findings signify the dynamic and reciprocal cross-talk between tumour cells and the metastatic niche; importantly, they provide new opportunities for effective anti-metastasis therapies, especially of consequence for brain metastasis patients.

Wnt-induced deubiquitination FoxM1 ensures nucleus ß-catenin transactivation.

A key step of Wnt signaling activation is the recruitment of ß-catenin to the Wnt target-gene promoter in the nucleus, but its mechanisms are largely unknown. Here, we identified FoxM1 as a novel target of Wnt signaling, which is essential for ß-catenin/TCF4 transactivation. GSK3 phosphorylates FoxM1 on serine 474 which induces FoxM1 ubiquitination mediated by FBXW7. Wnt signaling activation inhibits FoxM1 phosphorylation by GSK3-Axin complex and leads to interaction between FoxM1 and deubiquitinating enzyme USP5, thereby deubiquitination and stabilization of FoxM1. FoxM1 accumulation in the nucleus promotes recruitment of ß-catenin to Wnt target-gene promoter and activates the Wnt signaling pathway by protecting the ß-catenin/TCF4 complex from ICAT inhibition. Subsequently, the USP5-FoxM1 axis abolishes the inhibitory effect of ICAT and is required for Wnt-mediated tumor cell proliferation. Therefore, Wnt-induced deubiquitination of FoxM1 represents a novel and critical mechanism for controlling canonical Wnt signaling and cell proliferation.

Methylation of RNA N6-methyladenosine in modulation of cytokine responses and tumorigenesis.

Among myriads of distinct chemical modification in RNAs, the dynamic, reversible and fine-tuned methylation of N6-methyladenosine (m6A) is the most prevalent modification in eukaryotic mRNAs. This RNA mark is generated by proteins that act as m6A writers and can be reversed by proteins that act as m6A erasers. The RNA m6A modification is also mediated by another group of proteins capable of recognizing m6A that act as m6A readers. The m6A modification exerts direct control over the RNA metabolism including mRNA processing, mRNA exporting, translation initiation, mRNA stability and the biogenesis of long-non-coding RNA (LncRNA), thereby can influence various aspects of cell function. Evidently, m6A is intimately associated with cancer development and progression such as self-renewal capacity of cancer stem cells, proliferation, apoptosis and therapeutic resistance, and immune response. In this review, we will discuss the regulation and function of m6A, the various functions ascribed to these proteins and the emerging concepts that impact our knowledge of these proteins and their roles in the epitranscriptome. Conceivably, m6A may play pivotal roles in cytokine and immune response and carcinogenesis.

Gli1-induced deubiquitinase USP48 aids glioblastoma tumorigenesis by stabilizing Gli1.

Aberrant activation of the Hedgehog (Hh) signaling pathway drives the tumorigenesis of multiple cancers. In this study, we screened a panel of deubiquitinases that may regulate the Hh pathway. We find that deubiquitinase USP48 activates Gli-dependent transcription by stabilizing Gli1 protein. Mechanistically, USP48 interacts with Gli1 and cleaves its ubiquitin off directly. In glioblastoma cells, knockdown of USP48 inhibits cell proliferation and the expression of Gli1's downstream targets, which leads to repressed glioblastoma tumorigenesis. Importantly, USP48's effect on cell proliferation and tumorigenesis depends to some extent on Gli1. In addition, we find that the Sonic Hedgehog (SHH) pathway induces USP48 expression through Gli1-mediated transcriptional activation, which forms thus a positive feedback loop to regulate Hh signaling. In human glioblastoma specimens, the expression levels of USP48 and Gli1 proteins are clinically relevant, and high expression of USP48 correlates with glioma malignancy. In summary, our study reveals that the USP48-Gli1 regulatory axis is critical for glioma cell proliferation and glioblastoma tumorigenesis.

Sufficient dimension reduction via random-partitions for the large-p-small-n problem.

Sufficient dimension reduction (SDR) continues to be an active field of research. When estimating the central subspace (CS), inverse regression based SDR methods involve solving a generalized eigenvalue problem, which can be problematic under the large-p-small-n situation. In recent years, new techniques have emerged in numerical linear algebra, called randomized algorithms or random sketching, for high-dimensional and large scale problems. To overcome the large-p-small-n SDR problem, we combine the idea of statistical inference with random sketching to propose a new SDR method, called integrated random-partition SDR (iRP-SDR). Our method consists of the following three steps: (i) Randomly partition the covariates into subsets to construct an envelope subspace with low dimension. (ii) Obtain a sketch of the CS by applying a conventional SDR method within the constructed envelope subspace. (iii) Repeat the above two steps many times and integrate these multiple sketches to form the final estimate of the CS. After describing the details of these steps, the asymptotic properties of iRP-SDR are established. Unlike existing methods, iRP-SDR does not involve the determination of the structural dimension until the last stage, which makes it more adaptive to a high-dimensional setting. The advantageous performance of iRP-SDR is demonstrated via simulation studies and a practical example analyzing EEG data.

Profiles of brain metastases: Prioritization of therapeutic targets.

We sought to compare the tumor profiles of brain metastases from common cancers with those of primary tumors and extracranial metastases in order to identify potential targets and prioritize rational treatment strategies. Tumor samples were collected from both the primary and metastatic sites of nonsmall cell lung cancer, breast cancer and melanoma from patients in locations worldwide, and these were submitted to Caris Life Sciences for tumor multiplatform analysis, including gene sequencing (Sanger and next-generation sequencing with a targeted 47-gene panel), protein expression (assayed by immunohistochemistry) and gene amplification (assayed by in situ hybridization). The data analysis considered differential protein expression, gene amplification and mutations among brain metastases, extracranial metastases and primary tumors. The analyzed population included: 16,999 unmatched primary tumor and/or metastasis samples: 8,178 nonsmall cell lung cancers (5,098 primaries; 2,787 systemic metastases; 293 brain metastases), 7,064 breast cancers (3,496 primaries; 3,469 systemic metastases; 99 brain metastases) and 1,757 melanomas (660 primaries; 996 systemic metastases; 101 brain metastases). TOP2A expression was increased in brain metastases from all 3 cancers, and brain metastases overexpressed multiple proteins clustering around functions critical to DNA synthesis and repair and implicated in chemotherapy resistance, including RRM1, TS, ERCC1 and TOPO1. cMET was overexpressed in melanoma brain metastases relative to primary skin specimens. Brain metastasis patients may particularly benefit from therapeutic targeting of enzymes associated with DNA synthesis, replication and/or repair.

Robust mislabel logistic regression without modeling mislabel probabilities.

Logistic regression is among the most widely used statistical methods for linear discriminant analysis. In many applications, we only observe possibly mislabeled responses. Fitting a conventional logistic regression can then lead to biased estimation. One common resolution is to fit a mislabel logistic regression model, which takes into consideration of mislabeled responses. Another common method is to adopt a robust M-estimation by down-weighting suspected instances. In this work, we propose a new robust mislabel logistic regression based on γ-divergence. Our proposal possesses two advantageous features: (1) It does not need to model the mislabel probabilities. (2) The minimum γ-divergence estimation leads to a weighted estimating equation without the need to include any bias correction term, that is, it is automatically bias-corrected. These features make the proposed γ-logistic regression more robust in model fitting and more intuitive for model interpretation through a simple weighting scheme. Our method is also easy to implement, and two types of algorithms are included. Simulation studies and the Pima data application are presented to demonstrate the performance of γ-logistic regression.

Neuroprotective effects of adjunctive treatments for acute stroke thrombolysis: a review of clinical evidence.

The narrow therapeutic time window and risk of intracranial hemorrhage largely restrict the clinical application of thrombolysis in acute ischemic stroke. Adjunctive treatments added to rt-PA may be beneficial to improve the capacity of neural cell to withstand ischemia, and to reduce the hemorrhage risk as well. This review aims to evaluate the neuroprotective effects of adjunctive treatments in combination with thrombolytic therapy for acute ischemic stroke. Relevant studies were searched in the PubMed, Web of Science and EMBASE database. In this review, we first interpret the potential role of adjunctive treatments to thrombolytic therapy in acute ischemic stroke. Furthermore, we summarize the current clinical evidence for the combination of intravenous recombinant tissue plasminogen activator and various adjunctive therapies in acute ischemic stroke, either pharmacological or non-pharmacological therapy, and discuss the mechanisms of some promising treatments, including uric acid, fingolimod, minocycline, remote ischemic conditioning, hypothermia and transcranial laser therapy. Even though fingolimod, minocycline, hypothermia and remote ischemic conditioning have yielded promising results, they still need to be rigorously investigated in further clinical trials. Further trials should also focus on neuroprotective approach with pleiotropic effects or combined agents with multiple protective mechanisms.

Mir-21-Sox2 Axis Delineates Glioblastoma Subtypes with Prognostic Impact.

Glioblastoma (GBM) is the most aggressive human brain tumor. Although several molecular subtypes of GBM are recognized, a robust molecular prognostic marker has yet to be identified. Here, we report that the stemness regulator Sox2 is a new, clinically important target of microRNA-21 (miR-21) in GBM, with implications for prognosis. Using the MiR-21-Sox2 regulatory axis, approximately half of all GBM tumors present in the Cancer Genome Atlas (TCGA) and in-house patient databases can be mathematically classified into high miR-21/low Sox2 (Class A) or low miR-21/high Sox2 (Class B) subtypes. This classification reflects phenotypically and molecularly distinct characteristics and is not captured by existing classifications. Supporting the distinct nature of the subtypes, gene set enrichment analysis of the TCGA dataset predicted that Class A and Class B tumors were significantly involved in immune/inflammatory response and in chromosome organization and nervous system development, respectively. Patients with Class B tumors had longer overall survival than those with Class A tumors. Analysis of both databases indicated that the Class A/Class B classification is a better predictor of patient survival than currently used parameters. Further, manipulation of MiR-21-Sox2 levels in orthotopic mouse models supported the longer survival of the Class B subtype. The MiR-21-Sox2 association was also found in mouse neural stem cells and in the mouse brain at different developmental stages, suggesting a role in normal development. Therefore, this mechanism-based classification suggests the presence of two distinct populations of GBM patients with distinguishable phenotypic characteristics and clinical outcomes. SIGNIFICANCE STATEMENT: Molecular profiling-based classification of glioblastoma (GBM) into four subtypes has substantially increased our understanding of the biology of the disease and has pointed to the heterogeneous nature of GBM. However, this classification is not mechanism based and its prognostic value is limited. Here, we identify a new mechanism in GBM (the miR-21-Sox2 axis) that can classify ∼50% of patients into two subtypes with distinct molecular, radiological, and pathological characteristics. Importantly, this classification can predict patient survival better than the currently used parameters. Further, analysis of the miR-21-Sox2 relationship in mouse neural stem cells and in the mouse brain at different developmental stages indicates that miR-21 and Sox2 are predominantly expressed in mutually exclusive patterns, suggesting a role in normal neural development.

Forkhead Box M1 Is Essential for Nuclear Localization of Glioma-associated Oncogene Homolog 1 in Glioblastoma Multiforme Cells by Promoting Importin-7 Expression.

The transcription factors glioma-associated oncogene homolog 1 (GLI1), a primary marker of Hedgehog pathway activation, and Forkhead box M1 (FOXM1) are aberrantly activated in a wide range of malignancies, including glioma. However, the mechanism of nuclear localization of GLI1 and whether FOXM1 regulates the Hedgehog signaling pathway are poorly understood. Here we found that FOXM1 promotes nuclear import of GLI1 in glioblastoma multiforme cells and thus increases the expression of its target genes. Conversely, knockdown of FOXM1 expression with FOXM1 siRNA abrogated its nuclear import and inhibited the expression of its target genes. Also, genetic deletion of FOXM1 in mouse embryonic fibroblasts abolished nuclear localization of GLI1. We observed that FOXM1 directly binds to the importin-7 (IPO7) promoter and increases its promoter activity. IPO7 interacted with GLI1, leading to enhanced nuclear import of GLI1. Depletion of IPO7 by IPO7 siRNA reduced nuclear accumulation of GLI1. In addition, FOXM1 induced nuclear import of GLI1 by promoting IPO7 expression. Moreover, the FOXM1/IPO7/GLI1 axis promoted cell proliferation, migration, and invasion in vitro. Finally, expression of FOXM1 was markedly correlated with that of GLI1 in human glioblastoma specimens. These data suggest that FOXM1 and GLI1 form a positive feedback loop that contributes to glioblastoma development. Furthermore, our study revealed a mechanism that controls nuclear import of GLI1 in glioblastoma multiforme cells.

Detection of gene-gene interactions using multistage sparse and low-rank regression.

Finding an efficient and computationally feasible approach to deal with the curse of high-dimensionality is a daunting challenge faced by modern biological science. The problem becomes even more severe when the interactions are the research focus. To improve the performance of statistical analyses, we propose a sparse and low-rank (SLR) screening based on the combination of a low-rank interaction model and the Lasso screening. SLR models the interaction effects using a low-rank matrix to achieve parsimonious parametrization. The low-rank model increases the efficiency of statistical inference and, hence, SLR screening is able to more accurately detect gene-gene interactions than conventional methods. Incorporation of SLR screening into the Screen-and-Clean approach (Wasserman and Roeder, 2009; Wu et al., 2010) is also discussed, which suffers less penalty from Boferroni correction, and is able to assign p-values for the identified variables in high-dimensional model. We apply the proposed screening procedure to the Warfarin dosage study and the CoLaus study. The results suggest that the new procedure can identify main and interaction effects that would have been omitted by conventional screening methods.

Down-regulation of microRNA-494 via loss of SMAD4 increases FOXM1 and ß-catenin signaling in pancreatic ductal adenocarcinoma cells.

BACKGROUND & AIMS: Dysregulation of ß-catenin and the transcriptional activator FOXM1 mediate oncogenesis, but it is not clear how these proteins become dysregulated in tumors that do not typically carry mutations in adenomatous polyposis coli (APC) or ß-catenin, such as pancreatic ductal adenocarcinomas (PDACs). We searched for microRNAs that regulate levels of FOXM1 in PDAC cells and samples from patients. METHODS: We identified microRNAs that affect levels of FOXM1 in PDACs using bioinformatic, genetic, and pharmacologic approaches. We altered expression of the microRNA-494 (miR-494) in PDAC cell lines (AsPC-1 and PANC-1) and examined the effects on FOXM1 and ß-catenin signaling and cell proliferation and colony formation. The cells were injected into immunocompromised mice and growth of xenograft tumors and liver metastases were measured. We performed immunohistochemical analyses of 10 paired PDAC and nontumor pancreatic tissue samples collected from untreated patients during surgery. RESULTS: We identified miR-494 as a negative regulator of FOXM1 levels in PDAC cells, and found that levels of this microRNA were reduced in PDAC specimens, compared with nontumor tissues. Loss of response of PDAC cells to transforming growth factor ß, owing to SMAD4 deficiency, reduced expression of miR-494. Transgenic expression of miR-494 in PDAC cells produced the same effects as reducing expression of FOXM1 or blocking nuclear translocation of ß-catenin, reducing cell proliferation, migration, and invasion, and increasing their sensitivity to gemcitabine. Reduced expression of miR-494 correlated with PDAC metastasis and reduced survival times of patients. CONCLUSIONS: Loss of SMAD4 in PDAC cells leads to reduced levels of miR-494, increased levels of FOXM1, and nuclear localization of ß-catenin. miR-494 might be developed as a prognostic marker for patients with PDAC or a therapeutic target.

Elevated expression of HABP1 is a novel prognostic indicator in triple-negative breast cancers.

Hyaluronan-binding protein 1 (HABP1) has been documented to overexpress in several malignancies. The aim of our study is to investigate the expression of HABP1 protein in triple-negative breast cancer (TNBC) and its clinical significance. Using immunohistochemistry, HABP1 expression was evaluated in 139 TNBC specimens. The association between HABP1 expression with clinicopathological parameters was assessed using chi-square test. The survival status of patients was analyzed using the Kaplan-Meier and log-rank tests. Cox regression was used for the multivariate analysis of prognostic factors. The results showed that HABP1 overexpression correlated with higher histological grade (P = 0.004), advanced TNM stage (P = 0.008), greater tumor size (P = 0.016), metastasis of lymph node (P < 0.001), and recurrence (P = 0.040). Furthermore, it indicated that patients with HABP1 overexpression had significantly poorer overall survival (OS) and disease-free survival (DFS) compared with patients (P < 0.001 for both). Multivariate Cox regression analysis revealed that elevated HABP1 expression was an independent prognostic factor for both OS (P = 0.035) and DFS (P = 0.013). In contrast to the effect of HABP1 in TNBCs with lymph node metastasis, HABP1 overexpression significantly affects prognosis (OS, P = 0.002; DFS, P = 0.003) in TNBCs without lymph node metastasis. In conclusion, HABP1 protein high expression may contribute to the tumor progression and poor prognosis of TNBC, especially in predicting prognosis in TNBCs without lymph node metastasis.

Forkhead box M1 is regulated by heat shock factor 1 and promotes glioma cells survival under heat shock stress.

The forkhead box M1 (FoxM1) is a key transcription factor regulating multiple aspects of cell biology. Prior studies have shown that FoxM1 is overexpressed in a variety of human tumors, including brain tumor, and plays a critical role in cancer development and progression. In this study we found that FoxM1 was up-regulated by heat shock factor 1 (HSF1) under heat shock stress condition in multiple cell lines. Knockdown of HSF1 with HSF1 siRNA or inhibition of HSF1 with a HSF1 inhibitor abrogated heat shock-induced expression of FoxM1. Genetic deletion of HSF1 in mouse embryo fibroblast cells also abolished heat shock stress-induced FoxM1 expression. Moreover, we showed that HSF1 directly bound to FoxM1 promoter and increased FoxM1 promoter activity. Furthermore, we demonstrated that FoxM1 was required for the G(2)-M phase progression through regulating Cdc2, Cdc20, and Cdc25B under a mild heat shock stress but enhanced cell survival under lethal heat shock stress condition. Finally, in human glioblastoma specimens, FoxM1 overexpression correlated with elevated HSF1 expression. Our results indicate that FoxM1 is regulated by HSF1 and is critical for HSF1-mediated heat shock response. We demonstrated a novel mechanism of stress resistance controlled by HSF1 and a new HSF-FoxM1 connection that mediates cellular thermotolerance.

Disruption of Klf4 in villin-positive gastric progenitor cells promotes formation and progression of tumors of the antrum in mice.

BACKGROUND & AIMS: Krüppel-like factor 4 (Klf4) is a putative gastric tumor suppressor gene. Rare, villin-positive progenitor cells in the gastric antrum have multilineage potential. We investigated the function of Klf4 in these cells and in gastric carcinogenesis. METHODS: We created mice with disruption of Klf4 in villin-positive antral mucosa cells (Villin-Cre(+);Klf4(fl/fl) mice). Villin-Cre(+);Klf4(fl/fl) and control mice were given drinking water with or without 240 ppm N-methyl-N-nitrosourea at 5 weeks of age and thereafter on alternating weeks for a total of 10 weeks. Gastric mucosa samples were collected at 35, 50, or 80 weeks of age from mice that were and were not given N-methyl-N-nitrosourea, and analyzed by histopathologic and molecular analyses. Findings were compared with those from human gastric tumor specimens. RESULTS: Preneoplasia formed progressively in the antrum in 35- to 80-week-old Villin-Cre(+);Klf4(fl/fl) mice. Gastric tumors developed in 29% of 80-week-old Villin-Cre(+);Klf4(fl/fl) mice, which were located exclusively in the lesser curvature of the antrum. N-methyl-N-nitrosourea accelerated tumor formation, and tumors developed significantly more frequently in Villin-Cre(+);Klf4(fl/fl) mice than in control mice, at 35 and 50 weeks of age. Mouse and human gastric tumors had reduced expression of Krüppel-like factor 4 and increased expression of FoxM1 compared with healthy gastric tissue. Expression of Krüppel-like factor 4 suppressed transcription of FoxM1. CONCLUSIONS: Inactivation of Klf4 in villin-positive gastric progenitor cells induces transformation of the gastric mucosa and tumorigenesis in the antrum in mice. Villin-Cre(+);Klf4(fl/fl) have greater susceptibility to chemical-induced gastric carcinogenesis and increased rates of gastric tumor progression than control mice.

Dysregulated Krüppel-like factor 4 and vitamin D receptor signaling contribute to progression of hepatocellular carcinoma.

BACKGROUND & AIMS: Krüppel-like factor 4 (KLF4) is a transcription factor and putative tumor suppressor. However, little is known about its effects in hepatocellular carcinogenesis. We investigated the clinical significance, biologic effects, and mechanisms of dysregulated KLF4 signaling. METHODS: We performed microarray analysis of hepatocellular carcinoma (HCC) tissues. We used molecular biology analyses and animal models to evaluate activation and function of KLF4-vitamin D receptor (VDR) pathway. RESULTS: Expression of KLF4 protein was decreased or lost in primary HCC samples, in particular, lymph node metastases, compared with normal liver tissues. Loss of KLF4 from primary tumors was significantly associated with reduced survival time and was identified as a prognostic marker. Most human HCC cell lines had losses or substantial decreases in levels of KLF4. Exogenous expression of KLF4 in HCC cells upregulated expression of mesenchymal-epithelial transition (MET) and inhibited their migration, invasion, and proliferation in vitro. When these cells were injected into mice, tumors grew more slowly and metastasis was inhibited, compared with HCC cells that did not express KLF4. VDR is a direct transcriptional target of KLF4; we identified 2 sites in the VDR promoter that bound specifically to KLF4. Increased expression of VDR sensitized tumor cells to the inhibitory effects of vitamin D. CONCLUSIONS: KLF4 binds to the promoter of VDR to regulate its expression; levels of KLF4 are reduced and levels of VDR are increased in HCC cell lines and primary tumor samples. Expression of KLF4 in HCC cells sensitizes them to the anti-proliferative effects of VD3. This pathway might be manipulated to prevent or treat liver cancer.

Machine learning-based monosaccharide profiling for tissue-specific classification of Wolfiporia extensa samples.

Machine learning (ML) has been used for many clinical decision-making processes and diagnostic procedures in bioinformatics applications. We examined eight algorithms, including linear discriminant analysis (LDA), logistic regression (LR), k-nearest neighbor (KNN), random forest (RF), gradient boosting machine (GBM), support vector machine (SVM), Naïve Bayes classifier (NB), and artificial neural network (ANN) models, to evaluate their classification and prediction capabilities for four tissue types in Wolfiporia extensa using their monosaccharide composition profiles. All 8 ML-based models were assessed as exemplary models with AUC exceeding 0.8. Five models, namely LDA, KNN, RF, GBM, and ANN, performed excellently in the four-tissue-type classification (AUC > 0.9). Additionally, all eight models were evaluated as good predictive models with AUC value > 0.8 in the three-tissue-type classification. Notably, all 8 ML-based methods outperformed the single linear discriminant analysis (LDA) plotting method. For large sample sizes, the ML-based methods perform better than traditional regression techniques and could potentially increase the accuracy in identifying tissue samples of W. extensa.

Multiomics analyses reveal DARS1-AS1/YBX1-controlled posttranscriptional circuits promoting glioblastoma tumorigenesis/radioresistance.

The glioblastoma (GBM) stem cell-like cells (GSCs) are critical for tumorigenesis/therapeutic resistance of GBM. Mounting evidence supports tumor-promoting function of long noncoding RNAs (lncRNAs), but their role in GSCs remains poorly understood. By combining CRISPRi screen with orthogonal multiomics approaches, we identified a lncRNA DARS1-AS1-controlled posttranscriptional circuitry that promoted the malignant properties of GBM cells/GSCs. Depleting DARS1-AS1 inhibited the proliferation of GBM cells/GSCs and self-renewal of GSCs, prolonging survival in orthotopic GBM models. DARS1-AS1 depletion also impaired the homologous recombination (HR)-mediated double-strand break (DSB) repair and enhanced the radiosensitivity of GBM cells/GSCs. Mechanistically, DARS1-AS1 interacted with YBX1 to promote target mRNA binding and stabilization, forming a mixed transcriptional/posttranscriptional feed-forward loop to up-regulate expression of the key regulators of G1-S transition, including E2F1 and CCND1. DARS1-AS1/YBX1 also stabilized the mRNA of FOXM1, a master transcription factor regulating GSC self-renewal and DSB repair. Our findings suggest DARS1-AS1/YBX1 axis as a potential therapeutic target for sensitizing GBM to radiation/HR deficiency-targeted therapy.