Multi-task consensus clustering of genome-wide transcriptomes from related biological conditions

Screening and Comprehensive Analysis of Cancer-Associated tRNA-Derived Fragments

tRNA-derived fragments (tRFs) constitute a novel class of small non-coding RNA cleaved from tRNAs. In recent years, researches have shown the regulatory roles of a few tRFs in cancers, illuminating a new direction for tRF-centric cancer researches. Nonetheless, more specific screening of tRFs related to oncogenesis pathways, cancer progression stages and cancer prognosis is continuously demanded to reveal the landscape of the cancer-associated tRFs. In this work, by combining the clinical information recorded in The Cancer Genome Atlas (TCGA) and the tRF expression profiles curated by MINTbase v2.0, we systematically screened 1,516 cancer-associated tRFs (ca-tRFs) across seven cancer types. The ca-tRF set collectively combined the differentially expressed tRFs between cancer samples and control samples, the tRFs significantly correlated with tumor stage and the tRFs significantly correlated with patient survival. By incorporating our previous tRF-target dataset, we found the ca-tRFs tend to target cancer-associated genes and onco-pathways like ATF6-mediated unfolded protein response, angiogenesis, cell cycle process regulation, focal adhesion, PI3K-Akt signaling pathway, cellular senescence and FoxO signaling pathway across multiple cancer types. And cell composition analysis implies that the expressions of ca-tRFs are more likely to be correlated with T-cell infiltration. We also found the ca-tRF expression pattern is informative to prognosis, suggesting plausible tRF-based cancer subtypes. Together, our systematic analysis demonstrates the potentially extensive involvements of tRFs in cancers, and provides a reasonable list of cancer-associated tRFs for further investigations.

Identification of Gene-Set Signature in Early-Stage Hepatocellular Carcinoma and Relevant Immune Characteristics

Background

The incidence of hepatocellular carcinoma (HCC) is rising worldwide, and there is limited therapeutic efficacy due to tumor microenvironment heterogeneity and difficulty in early-stage screening. This study aimed to develop and validate a gene set-based signature for early-stage HCC (eHCC) patients and further explored specific marker dysregulation mechanisms as well as immune characteristics.

Methods

We performed an integrated bioinformatics analysis of genomic, transcriptomic, and clinical data with three independent cohorts. We systematically reviewed the crosstalk between specific genes, tumor prognosis, immune characteristics, and biological function in the different pathological stage samples. Univariate and multivariate survival analyses were performed in The Cancer Genome Atlas (TCGA) patients with survival data. Diethylnitrosamine (DEN)-induced HCC in Wistar rats was employed to verify the reliability of the predictions.

Results

We identified a Cluster gene that potentially segregates patients with eHCC from non-tumor, through integrated analysis of expression, overall survival, immune cell characteristics, and biology function landscapes. Immune infiltration analysis showed that lower infiltration of specific immune cells may be responsible for significantly worse prognosis in HCC (hazard ratio, 1.691; 95% CI: 1.171–2.441; p = 0.012), such as CD8 Tem and cytotoxic T cells (CTLs) in eHCC. Our results identified that Cluster C1 signature presented a high accuracy in predicting CD8 Tem and CTL immune cells (receiver operating characteristic (ROC) = 0.647) and cancerization (ROC = 0.946) in liver. As a central member of Cluster C1, overexpressed PRKDC was associated with the higher genetic alteration in eHCC than advanced-stage HCC (aHCC), which was also connected to immune cell-related poor prognosis. Finally, the predictive outcome of Cluster C1 and PRKDC alteration in DEN-induced eHCC rats was also confirmed.

Conclusions

As a tumor prognosis-relevant gene set-based signature, Cluster C1 showed an effective approach to predict cancerization of eHCC and its related immune characteristics with considerable clinical value.

A big data pipeline: Identifying dynamic gene regulatory networks from time-course Gene Expression Omnibus data with applications to influenza infection

A biological host response to an external stimulus or intervention such as a disease or infection is a dynamic process, which is regulated by an intricate network of many genes and their products. Understanding the dynamics of this gene regulatory network allows us to infer the mechanisms involved in a host response to an external stimulus, and hence aids the discovery of biomarkers of phenotype and biological function. In this article, we propose a modeling/analysis pipeline for dynamic gene expression data, called Pipeline4DGEData, which consists of a series of statistical modeling techniques to construct dynamic gene regulatory networks from the large volumes of high-dimensional time-course gene expression data that are freely available in the Gene Expression Omnibus repository. This pipeline has a consistent and scalable structure that allows it to simultaneously analyze a large number of time-course gene expression data sets, and then integrate the results across different studies. We apply the proposed pipeline to influenza infection data from nine studies and demonstrate that interesting biological findings can be discovered with its implementation.