In-depth cDNA library sequencing provides quantitative gene expression profiling in cancer biomarker discovery

RNA Biomarkers: Frontier of Precision Medicine for Cancer

As an essential part of central dogma, RNA delivers genetic and regulatory information and reflects cellular states. Based on high-throughput sequencing technologies, cumulating data show that various RNA molecules are able to serve as biomarkers for the diagnosis and prognosis of various diseases, for instance, cancer. In particular, detectable in various bio-fluids, such as serum, saliva and urine, extracellular RNAs (exRNAs) are emerging as non-invasive biomarkers for earlier cancer diagnosis, tumor progression monitor, and prediction of therapy response. In this review, we summarize the latest studies on various types of RNA biomarkers, especially extracellular RNAs, in cancer diagnosis and prognosis, and illustrate several well-known RNA biomarkers of clinical utility. In addition, we describe and discuss general procedures and issues in investigating exRNA biomarkers, and perspectives on utility of exRNAs in precision medicine.

A robust hybrid approach based on estimation of distribution algorithm and support vector machine for hunting candidate disease genes

Microarray data are high dimension with high noise ratio and relatively small sample size, which makes it a challenge to use microarray data to identify candidate disease genes. Here, we have presented a hybrid method that combines estimation of distribution algorithm with support vector machine for selection of key feature genes. We have benchmarked the method using the microarray data of both diffuse B cell lymphoma and colon cancer to demonstrate its performance for identifying key features from the profile data of high-dimension gene expression. The method was compared with a probabilistic model based on genetic algorithm and another hybrid method based on both genetics algorithm and support vector machine. The results showed that the proposed method provides new computational strategy for hunting candidate disease genes from the profile data of disease gene expression. The selected candidate disease genes may help to improve the diagnosis and treatment for diseases.

Revealing stable processing products from ribosome-associated small RNAs by deep-sequencing data analysis

The exploration of the non-protein-coding RNA (ncRNA) transcriptome is currently focused on profiling of microRNA expression and detection of novel ncRNA transcription units. However, recent studies suggest that RNA processing can be a multi-layer process leading to the generation of ncRNAs of diverse functions from a single primary transcript. Up to date no methodology has been presented to distinguish stable functional RNA species from rapidly degraded side products of nucleases. Thus the correct assessment of widespread RNA processing events is one of the major obstacles in transcriptome research. Here, we present a novel automated computational pipeline, named APART, providing a complete workflow for the reliable detection of RNA processing products from next-generation-sequencing data. The major features include efficient handling of non-unique reads, detection of novel stable ncRNA transcripts and processing products and annotation of known transcripts based on multiple sources of information. To disclose the potential of APART, we have analyzed a cDNA library derived from small ribosome-associated RNAs in Saccharomyces cerevisiae. By employing the APART pipeline, we were able to detect and confirm by independent experimental methods multiple novel stable RNA molecules differentially processed from well known ncRNAs, like rRNAs, tRNAs or snoRNAs, in a stress-dependent manner.