Introducing a Stable Bootstrap Validation Framework for Reliable Genomic Signature Extraction

A constraint score guided meta-heuristic searching to attribute reduction

Essentially, the problem solving of attribute reduction can be regarded as a process of reduct searching which will be terminated if a pre-defined restriction is achieved. Presently, among a variety of searching strategies, meta-heuristic searching has been widely accepted. Nevertheless, it should be emphasized that the iterative procedures in most meta-heuristic algorithms rely heavily on the random generation of initial population, such a type of generation is naturally associated with the limitations of inferior stability and performance. Therefore, a constraint score guidance is proposed before carrying out meta-heuristic searching and then a novel framework to seek out reduct is developed. Firstly, for each attribute and each label in data, the index called local constraint score is calculated. Secondly, the qualified attributes are identified by those constraint scores, which consist of the foundation of initial population. Finally, the meta-heuristic searching can be further employed to achieve the required restriction in attribute reduction. Note that most existing meta-heuristic searchings and popular measures (evaluate the significance of attributes) can be embedded into our framework. Comprehensive experiments over 20 public datasets clearly validated the effectiveness of our framework: it is beneficial to reduct with superior stabilities, and the derived reduct may further contribute to the improvement of classification performance.

Clinical outcomes and potential therapies prediction of subgroups based on a ferroptosis-related long non-coding RNA signature for gastric cancer

Background: Gastric cancer (GC) is one of the most aggressive malignant tumors worldwide. Ferroptosis is a kind of iron-dependent cell death, which is proved to be closely related to tumor progression. In this study, we aim at constructing a ferroptosis-related lncRNAs signature to predict the prognosis of GC and explore potential therapies.

Methods: Ferroptosis-Related LncRNAs Signature for GC patients (FRLSG) was constructed through univariate Cox regression, the LASSO algorithm, and multivariate Cox regression. Kaplan–Meier analysis, receiver operating characteristic curves, and risk score plot were applied to verify the predictive power of FRLSG. Gene Set Enrichment Analysis (GSEA) and immune infiltration analyses were conducted to explore the potential clinical value of the FRLSG. In addition, drug sensitivity prediction was applied to identify chemotherapeutic drugs with potential therapeutic effect.

Results: Five ferroptosis-related lncRNAs (AC004816.1, AC005532.1, LINC01357, AL355574.1 and AL049840.4) were identified to construct FRLSG, whose expression level in GC were confirmed by experimental validation. Kaplan-Meier curve and ROC curve proved the reliability and effectiveness of the FRLSG in predicting the prognosis for GC patients. Several immune-related pathways were enriched in the high-FRLSG group, and further immune infiltration analyses demonstrated the high immune infiltration status of the high-FRLSG group. In addition, 19 and 24 candidate drugs with potential therapeutic effect were identified for the high- and low-FRLSG groups, respectively.

Conclusions: FRLSG was an effective tool in predicting the prognosis of GC, which might help to prioritize potential therapeutics for GC patients.

Development and Validation of a 6-Gene Hypoxia-Related Prognostic Signature For Cholangiocarcinoma

Cholangiocarcinoma (CHOL) is highly malignant and has a poor prognosis. This study is committed to creating a new prognostic model based on hypoxia related genes. Here, we established a novel tumor hypoxia-related prognostic model consisting of 6 hypoxia-related genes by univariate Cox regression and the least absolute shrinkage and selection operator (LASSO) algorithm to predict CHOL prognosis and then the risk score for each patient was calculated. The results showed that the patients with high-risk scores had poor prognosis compared with those with low-risk scores, which was verified as an independent predictor by multivariate analysis. The hypoxia-related prognostic model was validated in both TCGA and GEO cohorts and exhibited excellent performance in predicting overall survival in CHOL. The PPI results suggested that hypoxia-related genes involved in the model may play a central role in regulating the hypoxic state. In addition, the presence of IDH1 mutations in the high-risk group was high, and GSEA results showed that some metabolic pathways were upregulated, but immune response processes were generally downregulated. These factors may be potential reasons for the high-risk group with worse prognosis. The analysis of different immune regulation-related processes in the high- and low-risk groups revealed that the expression of genes related to immune checkpoints would show differences between these two groups. We further verified the expression of the oncogene PPFIA4 in the model, and found that compared with normal samples, CHOL patients were generally highly expressed, and the patients with high-expression of PPFIA4 had a poor prognosis. In summary, the present study may provide a valid prognostic model for bile duct cancer to inform better clinical management of patients.

Development and validation of a survival model for esophageal adenocarcinoma based on autophagy-associated genes

Autophagy is a highly conserved catabolic process which has been implicated in esophageal adenocarcinoma (EAC). We sought to investigate the biological functions and prognostic value of autophagy-related genes (ARGs) in EAC. A total of 21 differentially expressed ARGs were identified between EAC and normal samples. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis were then applied for the differentially expressed ARGs in EAC, and the protein-protein interaction (PPI) network was established. Cox survival analysis and Lasso regression analysis were performed to establish a prognostic prediction model based on nine overall survival (OS)-related ARGs (CAPN1, GOPC, TBK1, SIRT1, ARSA, BNIP1, ERBB2, NRG2, PINK1). The 9-gene prognostic signature significantly stratified patient outcomes in The Cancer Genome of Atlas (TCGA)-EAC cohort and was considered as an independently prognostic predictor for EAC patients. Moreover, Gene set enrichment analysis (GSEA) analyses revealed several important cellular processes and signaling pathways correlated with the high-risk group in EAC. This prognostic prediction model was confirmed in an independent validation cohort (GSE13898) from The Gene Expression Omnibus (GEO) database. We also developed a nomogram with a concordance index of 0.78 to predict the survival possibility of EAC patients by integrating the risk signature and clinicopathological features. The calibration curves substantiated favorable concordance between actual observation and nomogram prediction. Last but not least, Erb-B2 Receptor Tyrosine Kinase 2 (ERBB2), a member of the prognostic gene signature, was identified as a potential therapeutic target for EAC patients. To sum up, we established and verified a novel prognostic prediction model based on ARGs which could optimize the individualized survival prediction in EAC.

Identification and validation of a hypoxia-related prognostic signature in clear cell renal cell carcinoma patients

Increasing evidence has shown that hypoxia is closely related to the development, progression, and prognosis of clear cell renal cell carcinoma (ccRCC). Nevertheless, reliable prognostic signatures based on hypoxia have not been well-established. This study aimed to establish a hypoxia-related prognostic signature and construct an optimized nomogram for patients with ccRCC.We accessed hallmark gene sets of hypoxia, including 200 genes, and an original RNA seq dataset of ccRCC cases with integrated clinical information obtained by mining the Cancer Genome Atlas database and the International Cancer Genome Consortium (ICGC) database. Univariate Cox regression analysis and multivariate Cox proportional hazards regression were performed to identify prognostic hub genes and further established prognostic model as well as visualized the nomogram. External validation of the optimized nomogram was performed in independent cohorts from the ICGC database.ANKZF1, ETS1, PLAUR, SERPINE1, FBP1, and PFKP were selected as prognostic hypoxia-related hub genes, and the prognostic model effectively distinguishes high-risk and low-risk patients with ccRCC. The results of receiver operating characteristic curve, risk plots, survival analysis, and independent analysis suggested that RiskScore was a useful tool and independent predictive factor. A novel prognosis nomogram optimized via RiskScore showed its promising performance in both the Cancer Genome Atlas-ccRCC cohort and an ICGC-ccRCC cohort.Our study reveals that the differential expressions of hypoxia-related genes are associated with the overall survival of patients with ccRCC. The prognostic model we established showed a good predictive and discerning ability in ccRCC patients. The novel nomogram optimized via RiskScore exhibited a promising predictive ability. It may be able to serve as a visualized tool for guiding clinical decisions and selecting effective individualized treatments.

Differential Expression of the TLR4 Gene in Pan-Cancer and Its Related Mechanism

Previous studies have revealed the relationship between toll-like receptor 4 (TLR4) polymorphisms and cancer susceptibility. However, the relationship between TLR4 and prognosis and immune cell infiltration in pan-cancer patients is still unclear. Through the Genotype-Tissue Expression (GTEx) and The Cancer Genome Atlas (TCGA) databases, the distinct expression of the TLR4 gene in 24 tumors and normal tissues was analyzed. Univariate Cox proportional hazards regression analysis was used to identify the cancer types whose TLR4 gene expression was related to prognosis. The relationship between TLR4 and tumor cell immune invasion was studied. Spearman's rank correlation coefficient was used to analyze the relationship among TLR4 and immune neoantigens, tumor mutation burden (TMB), microsatellite instability (MSI), DNA repair genes, and DNA methylation. Gene Set Enrichment Analysis (GSEA) was used to identify the tumor-related pathways that the TLR4 gene was highly expressed in; the expression of the TLR4 gene was verified with the Human Protein Atlas (HPA) database. Low expression of TLR4 was associated with an inferior prognosis in kidney renal clear cell carcinoma (KIRC), skin cutaneous melanoma (SKCM), and uterine corpus endometrial carcinoma (UCEC), while high expression was related to a poor prognosis in head and neck squamous cell carcinoma (HNSC), prostate adenocarcinoma (PRAD), stomach adenocarcinoma (STAD), and testicular germ cell tumor (TGCT). The expression of TLR4 was negatively correlated with the expression of B cells in STAD. The expression of TLR4 was positively correlated with the infiltration of B cells, CD4 and CD8 T cells, neutrophils, macrophages, and dendritic cells in STAD, KIRC, UCEC, TGCT, and SKCM. The expression of the TLR4 gene in KIRC, SKCM, STAD, TGCT, and UCEC was highly correlated with inducible T-cell costimulator (ICOS), cytotoxic T lymphocyte-associated molecule 4 (CTLA4), and CD28 immune checkpoints. Spearman's rank correlation coefficient showed that the expression of TLR4 gene was significantly correlated with TMB in STAD and UCEC and was prominently correlated with MSI in TGCT, STAD, and SKCM. The expression of the TLR4 gene was highly correlated with MLH1, MSH2, and MSH6 in KIRC, SKCM, and STAD. The expression of the TLR4 gene was remarkably correlated with the methyltransferases DNA methyltransferase 2 (DNMT2) and DNA methyltransferase 3-beta (DNMT3B) in SKCM and STAD. Enrichment analysis showed that TLR4 was highly expressed in the chemokine signaling pathway and the cell adhesion molecule and cytokine receptor interaction pathway. In summary, the expression of TLR4 is linked to the prognosis of KIRC, SKCM, STAD, TGCT, and UCEC patients and the level of immune infiltration of CD4, CD8 T cells, macrophages, neutrophils, and dendritic cells.

Application of an Autophagy-Related Gene Prognostic Risk Model Based on TCGA Database in Cervical Cancer

Background: Autophagy plays an important role in the development of cancer. However, the prognostic value of autophagy-related genes (ARGs) in cervical cancer (CC) is unclear. The purpose of this study is to construct a survival model for predicting the prognosis of CC patients based on ARG signature.

Methods: ARGs were obtained from the Human Autophagy Database and Molecular Signatures Database. The expression profiles of ARGs and clinical data were downloaded from the TCGA database. Differential expression analysis of CC tissues and normal tissues was performed using R software to screen out ARGs with an aberrant expression. Univariate Cox, Lasso, and multivariate Cox regression analyses were used to construct a prognostic model which was validated by using the test set and the entire set. We also performed an independent prognostic analysis of risk score and some clinicopathological factors of CC. Finally, a clinical practical nomogram was established to predict individual survival probability.

Results: Compared with normal tissues, there were 63 ARGs with an aberrant expression in CC tissues. A risk model based on 3 ARGs was finally obtained by Lasso and Cox regression analysis. Patients with high risk had significantly shorter overall survival (OS) than low-risk patients in both train set and validation set. The ROC curve validated its good performance in survival prediction, suggesting that this model has a certain extent sensitivity and specificity. Multivariate Cox analysis showed that the risk score was an independent prognostic factor. Finally, we mapped a nomogram to predict 1-, 3-, and 5-year survival for CC patients. The calibration curves indicated that the model was reliable.

Conclusion: A risk prediction model based on CHMP4C, FOXO1, and RRAGB was successfully constructed, which could effectively predict the prognosis of CC patients. This model can provide a reference for CC patients to make precise treatment strategy.

A sparse deep learning approach for automatic segmentation of human vasculature in multispectral optoacoustic tomography

Multispectral Optoacoustic Tomography (MSOT) resolves oxy- (HbO2) and deoxy-hemoglobin (Hb) to perform vascular imaging. MSOT suffers from gradual signal attenuation with depth due to light-tissue interactions: an effect that hinders the precise manual segmentation of vessels. Furthermore, vascular assessment requires functional tests, which last several minutes and result in recording thousands of images. Here, we introduce a deep learning approach with a sparse-UNET (S-UNET) for automatic vascular segmentation in MSOT images to avoid the rigorous and time-consuming manual segmentation. We evaluated the S-UNET on a test-set of 33 images, achieving a median DICE score of 0.88. Apart from high segmentation performance, our method based its decision on two wavelengths with physical meaning for the task-at-hand: 850 nm (peak absorption of oxy-hemoglobin) and 810 nm (isosbestic point of oxy-and deoxy-hemoglobin). Thus, our approach achieves precise data-driven vascular segmentation for automated vascular assessment and may boost MSOT further towards its clinical translation.

Identification and validation of an individualized autophagy-clinical prognostic index in gastric cancer patients

Background

The purpose of this study is to perform bioinformatics analysis of autophagy-related genes in gastric cancer, and to construct a multi-gene joint signature for predicting the prognosis of gastric cancer.

Methods

GO and KEGG analysis were applied for differentially expressed autophagy-related genes in gastric cancer, and PPI network was constructed in Cytoscape software. In order to optimize the prognosis evaluation system of gastric cancer, we established a prognosis model integrating autophagy-related genes. We used single factor Cox proportional risk regression analysis to screen genes related to prognosis from 204 autophagy-related genes in The Atlas Cancer Genome (TCGA) gastric cancer cohort. Then, the generated genes were applied to the Least Absolute Shrinkage and Selection Operator (LASSO). Finally, the selected genes were further included in the multivariate Cox proportional hazard regression analysis to establish the prognosis model. According to the median risk score, patients were divided into high-risk group and low-risk group, and survival analysis was conducted to evaluate the prognostic value of risk score. Finally, by combining clinic-pathological features and prognostic gene signatures, a nomogram was established to predict individual survival probability.

Results

GO analysis showed that the 28 differently expressed autophagy-related genes was enriched in cell growth, neuron death, and regulation of cell growth. KEGG analysis showed that the 28 differently expressed autophagy-related genes were related to platinum drug resistance, apoptosis and p53 signaling pathway. The risk score was constructed based on 4 genes (GRID2, ATG4D,GABARAPL2, CXCR4), and gastric cancer patients were significantly divided into high-risk and low-risk groups according to overall survival. In multivariate Cox regression analysis, risk score was still an independent prognostic factor (HR = 1.922, 95% CI = 1.573–2.349, P < 0.001). Cumulative curve showed that the survival time of patients with low-risk score was significantly longer than that of patients with high-risk score (P < 0.001). The external data GSE62254 proved that nomograph had a great ability to evaluate the prognosis of individual gastric cancer patients.

Conclusions

This study provides a potential prognostic marker for predicting the prognosis of GC patients and the molecular biology of GC autophagy.