AdaReg: data adaptive robust estimation in linear regression with application in GTEx gene expressions

The Genotype-Tissue Expression (GTEx) project provides a valuable resource of large-scale gene expressions across multiple tissue types. Under various technical noise and unknown or unmeasured factors, how to robustly estimate the major tissue effect becomes challenging. Moreover, different genes exhibit heterogeneous expressions across different tissue types. Therefore, we need a robust method which adapts to the heterogeneities of gene expressions to improve the estimation for the tissue effect. We followed the approach of the robust estimation based on γ-density-power-weight in the works of Fujisawa, H. and Eguchi, S. (2008). Robust parameter estimation with a small bias against heavy contamination. J. Multivariate Anal. 99: 2053–2081 and Windham, M.P. (1995). Robustifying model fitting. J. Roy. Stat. Soc. B: 599–609, where γ is the exponent of density weight which controls the balance between bias and variance. As far as we know, our work is the first to propose a procedure to tune the parameter γ to balance the bias-variance trade-off under the mixture models. We constructed a robust likelihood criterion based on weighted densities in the mixture model of Gaussian population distribution mixed with unknown outlier distribution, and developed a data-adaptive γ-selection procedure embedded into the robust estimation. We provided a heuristic analysis on the selection criterion and found that our practical selection trend under various γ’s in average performance has similar capability to capture minimizer γ as the inestimable mean squared error (MSE) trend from our simulation studies under a series of settings. Our data-adaptive robustifying procedure in the linear regression problem (AdaReg) showed a significant advantage in both simulation studies and real data application in estimating tissue effect of heart samples from the GTEx project, compared to the fixed γ procedure and other robust methods. At the end, the paper discussed some limitations on this method and future work.

Food products identified as source of a foodborne disease outbreak by a fast and robust likelihood estimation

Identifying a specific product causing a foodborne disease outbreak can be difficult, especially when dealing with a large amounts of suspicious food items and weak epidemiological evidence. A previously described likelihood model (Norström et al. 2015), improved within the OHEJP NOVA project, helps to prioritize food products that should be sampled for laboratory analysis. It is the aim of our study to integrate this approach into state of the art tracing software FoodChain-Lab (FCL; https://foodrisklabs.bfr.bund.de/foodchain-lab) developed at BfR to facilitate outbreak investigations. The model improved by Kausrud et al. in R (Ihaka and Gentleman 1996) uses wholesale data, the distribution of disease cases and census data to sort food items by their estimated likelihood to be the source of an outbreak. We developed a fast and secure intuitive software module using the Web Assembly technology (Haas et al. 2017) allowing professionals to embed the module easily into other applications. We integrated the module into the FCL web application for tracing (FCL Web; https://fcl-portal.bfr.berlin) to provide an intuitive and user-friendly solution. This solution combines a simple data input with extended data wrangling to make the calculation of the NOVA model as easy as possible. Since the model can be executed directly inside the web browser and therefore does not rely on any server environment, the possibility of data leakage can be highly reduced. The implementation of the advanced likelihood model into FCL Web increase the availability of this model and provides investigators easy, fast and reliable usage to improve outbreak investigation workflows.

A Glycolysis-Related Five-Gene Signature Predicts Biochemical Recurrence-Free Survival in Patients With Prostate Adenocarcinoma

Prostate cancer (PCa) is one of the most frequently diagnosed cancers in males worldwide. Approximately 25% of all patients experience biochemical recurrence (BCR) after radical prostatectomy (RP) and BCR indicates increased risk for metastasis and castration resistance. PCa patients with highly glycolytic tumors have a worse prognosis. Thus, this study aimed to explore glycolysis-based predictive biomarkers for BCR. Expression data and clinical information of PCa samples were retrieved from three publicly available datasets. One from The Cancer Genome Atlas (TCGA) dataset was used as the training cohort, and two from the Gene Expression Omnibus (GEO) dataset (GSE54460 and GSE70769) were used as validation cohorts. Using the training cohort, univariate Cox regression survival analysis, robust likelihood-based survival model, and stepwise multiply Cox analysis were sequentially applied to explore predictive glycolysis-related candidates. A five-gene risk score was then constructed based on the Cox coefficient as the following: (−0.8367*GYS2) + (0.3448*STMN1) + (0.3595*PPFIA4) + (−0.1940*KDELR3) + (0.4779*ABCB6). Receiver operating characteristic curve (ROC) analysis was used to identify the optimal cut-off point, and patients were divided into low risk and high risk groups. Kaplan–Meier analysis revealed that high risk group had significantly shorter BCR free survival time as compared with that in low risk group in training and validation cohorts. In conclusion, our data support the glycolysis-based five-gene signature as a novel and robust signature for predicting BCR of PCa patients.

On the determinants, gains and challenges of electronic banking adoption in Nigeria

PurposeThe purpose of this study is to examine the gains, challenges and determinants of electronic banking adoption in Nigeria.Design/methodology/approachThis paper applied the generalized structural equation modelling (GSEM) to a large sample of respondents surveyed from five of the six geopolitical zones of Nigeria to model the determinants of electronic banking. In addition to many other advantages, GSEM can be used as a likelihood function. As a result, this paper proposes GSEM as the most appropriate tool for modelling the socioeconomic determinant of electronic banking adoption.FindingsAbout three-quarter of respondents adopted at least a form of electronic banking. However, only a tenth of users used e-banking for purchase of goods or services, implying low electronic payment adoption. The low adoption of electronic payment was due to poor digital security infrastructure which made users vulnerable to widespread electronic frauds. The findings also show that the adoption of e-banking platforms or services was characterized by users' socioeconomic status. For example, the odds of adopting internet/mobile banking decreases with older users but increase with higher educational attainment and income, whereas the odds of adopting e-banking platforms such as short message service (SMS) and point of sale (POS) banking increases with older users and informally employed users respectively.Practical implicationsFor a sustainable cashless economy and financial inclusion in Nigeria, policy consolidation that provides safe e-banking services is necessary. Also, e-banking service providers should deliver specific contents and services that match the physical and economic characteristics of users.Originality/valueGeneralized structural equation modelling (GSEM) is a robust likelihood function method that combines the power of structural equation modelling with the generalized linear model. The application of GSEM to predict the likelihood of adopting a banking technology or Service has not been explored in electronic banking literature. Also, as a fast-growing economy with a heterogeneous population, Nigeria presents an interesting context to study the determinants of electronic banking.

A Ten-N6-Methyladenosine (m6A)-Modified Gene Signature Based on a Risk Score System Predicts Patient Prognosis in Rectum Adenocarcinoma

ObjectivesThe study aims to analyze the expression of N6-methyladenosine (m6A)-modified genes in rectum adenocarcinoma (READ) and identify reliable prognostic biomarkers to predict the prognosis of READ.Materials and MethodsRNA sequence data of READ and corresponding clinical survival data were obtained from The Cancer Genome Atlas (TCGA) database. N6-methyladenosine (m6A)-modified genes in READ were downloaded from the “m6Avar” database. Differentially expressed m6A-modified genes in READ stratified by different clinicopathological characteristics were identified using the “limma” package in R. Protein-protein interaction (PPI) network and co-expression analysis of differentially expressed genes (DEGs) were performed using “STRING” and Cytoscape, respectively. Principal component analysis (PCA) was done using R. In addition, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were used to functionally annotate the differentially expressed genes in different subgroups. Univariate Cox regression analyses were conducted to identify the powerful independent prognostic factors in READ associated with overall survival (OS). A robust likelihood-based survival model was built using the “rbsurv” package to screen for survival-associated signature genes. The Support Vector Machine (SVM) was used to predict the prognosis of READ through the risk score of survival-associated signature genes. Correlation analysis were carried out using GraphPad prism 8.ResultsWe screened 974 differentially expressed m6A-modified genes among four types of READ samples. Two READ subgroups (group 1 and group 2) were identified by K means clustering according to the expression of DEGs. The two subgroups were significantly different in overall survival and pathological stages. Next, 118 differentially expressed genes between the two subgroups were screened and the expression of 112 genes was found to be related to the prognosis of READ. Next, a panel of 10 survival-associated signature genes including adamtsl1, csmd2, fam13c, fam184a, klhl4, olfml2b, pdzd4, sec14l5, setbp1, tmem132b was constructed. The signature performed very well for prognosis prediction, time-dependent receiver-operating characteristic (ROC) analysis displaying an area under the curve (AUC) of 0.863, 0.8721, and 0.8752 for 3-year survival rate, prognostic status, and pathological stage prediction, respectively. Correlation analysis showed that the expression levels of the 10 m6A-modified genes were positively correlated with that of m6A demethylase FTO and ALKBH5.ConclusionThis study identified potential m6A-modified genes that may be involved in the pathophysiology of READ and constructed a novel gene expression panel for READ risk stratification and prognosis prediction.

COVID-19 case fatality risk by age and gender in a high testing setting in Latin America: Chile, March–August 2020

Background Early severity estimates of coronavirus disease 2019 (COVID-19) are critically needed to assess the potential impact of the ongoing pandemic in different demographic groups. Here we estimate the real-time delay-adjusted case fatality rate across nine age groups by gender in Chile, the country with the highest testing rate for COVID-19 in Latin America. Methods We used a publicly available real-time daily series of age-stratified COVID-19 cases and deaths reported by the Ministry of Health in Chile from the beginning of the epidemic in March through August 31, 2020. We used a robust likelihood function and a delay distribution to estimate real-time delay-adjusted case-fatality risk and estimate model parameters using a Monte Carlo Markov Chain in a Bayesian framework. Results As of August 31, 2020, our estimates of the time-delay adjusted case fatality rate (CFR) for men and women are 4.16% [95% Credible Interval (CrI): 4.09–4.24%] and 3.26% (95% CrI: 3.19–3.34%), respectively, while the overall estimate is 3.72% (95% CrI: 3.67–3.78%). Seniors aged 80 years and over have an adjusted CFR of 56.82% (95% CrI: 55.25–58.34%) for men and 41.10% (95% CrI: 40.02–42.26%) for women. Results showed a peak in estimated CFR during the June peak of the epidemic. The peak possibly reflects insufficient laboratory capacity, as illustrated by high test positivity rates (33% positive 7-day average nationally in June), which may have resulted in lower reporting rates. Conclusions Severity estimates from COVID-19 in Chile suggest that male seniors, especially among those aged ≥ 70 years, are being disproportionately affected by the pandemic, a finding consistent with other regions. The ongoing pandemic is imposing a high death toll in South America, and Chile has one of the highest reported mortality rates globally thus far. These real-time estimates may help inform public health officials' decisions in the region and underscore the need to implement more effective measures to ameliorate fatality.

A gene-based risk score model for predicting recurrence-free survival in patients with hepatocellular carcinoma

Background Hepatocellular carcinoma (HCC) remains the most frequent liver cancer, accounting for approximately 90% of primary liver cancers worldwide. The recurrence-free survival (RFS) of HCC patients is a critical factor in devising a personal treatment plan. Thus, it is necessary to accurately forecast the prognosis of HCC patients in clinical practice. Methods Using The Cancer Genome Atlas (TCGA) dataset, we identified genes associated with RFS. A robust likelihood-based survival modeling approach was used to select the best genes for the prognostic model. Then, the GSE76427 dataset was used to evaluate the prognostic model’s effectiveness. Results We identified 1331 differentially expressed genes associated with RFS. Seven of these genes were selected to generate the prognostic model. The validation in both the TCGA cohort and GEO cohort demonstrated that the 7-gene prognostic model can predict the RFS of HCC patients. Meanwhile, the results of the multivariate Cox regression analysis showed that the 7-gene risk score model could function as an independent prognostic factor. In addition, according to the time-dependent ROC curve, the 7-gene risk score model performed better in predicting the RFS of the training set and the external validation dataset than the classical TNM staging and BCLC. Furthermore, these seven genes were found to be related to the occurrence and development of liver cancer by exploring three other databases. Conclusion Our study identified a seven-gene signature for HCC RFS prediction that can be used as a novel and convenient prognostic tool. These seven genes might be potential target genes for metabolic therapy and the treatment of HCC.

Identification of the 11-lncRNA signatures associated with the prognosis of endometrial carcinoma

Endometrial carcinoma (EC) is the fourth most common cancer in women. Some long non-coding RNAs (lncRNAs) are regarded as potential prognostic biomarkers or targets for treatment of many types of cancers. We aim to screen prognostic-related lncRNAs and build a possible lncRNA signature which can effectively predict the survival of patients with EC. We obtained lncRNA expression profiling from the TCGA database. The patients were classified into training set and verification set. By performing Univariate Cox regression model, Robust likelihood-based survival analysis, and Cox proportional hazards model, we developed a risk score with the Cox co-efficient of individual lncRNAs in the training set. The optimum cut-off point was selected by ROC analysis. Patients were effectively divided into high-risk group and low-risk group according to the risk score. The OS of the low-risk patients was significantly prolonged compared with that of the high-risk group. At last, we validated this 11-lncRNA signature in the verification set and the complete set. We identified an 11-lncRNA expression signature with high stability and feasibility, which can predict the survival of patients with EC. These findings provide new potential biomarkers to improve the accuracy of prognosis prediction of EC.

Construction and validation of a 6-gene nomogram discriminating lung metastasis risk of breast cancer patients

Breast cancer is the most common malignant disease in women. Metastasis is the foremost cause of death. Breast tumor cells have a proclivity to metastasize to specific organs. The lung is one of the most common sites of breast cancer metastasis. Therefore, we aimed to build a useful and convenient prediction tool based on several genes that may affect lung metastasis-free survival (LMFS). We preliminarily identified 319 genes associated with lung metastasis in the training set GSE5327 (n = 58). Enrichment analysis of GO functions and KEGG pathways was conducted based on these genes. The best genes for modeling were selected using a robust likelihood-based survival modeling approach: GOLGB1, TMEM158, CXCL8, MCM5, HIF1AN, and TSPAN31. A prognostic nomogram for predicting lung metastasis in breast cancer was developed based on these six genes. The effectiveness of the nomogram was evaluated in the training set GSE5327 and the validation set GSE2603. Both the internal validation and the external validation manifested the effectiveness of our 6-gene prognostic nomogram in predicting the lung metastasis risk of breast cancer patients. On the other hand, in the validation set GSE2603, we found that neither the six genes in the nomogram nor the risk predicted by the nomogram were associated with bone metastasis of breast cancer, preliminarily suggesting that these genes and nomogram were specifically associated with lung metastasis of breast cancer. What’s more, five genes in the nomogram were significantly differentially expressed between breast cancer and normal breast tissues in the TIMER database. In conclusion, we constructed a new and convenient prediction model based on 6 genes that showed practical value in predicting the lung metastasis risk for clinical breast cancer patients. In addition, some of these genes could be treated as potential metastasis biomarkers for antimetastatic therapy in breast cancer. The evolution of this nomogram will provide a good reference for the prediction of tumor metastasis to other specific organs.