Cancer Transcriptomics and Epigenomics

​

​

​Gene Expression Profiling in Cancer

  Cancer, the disease that causes the greatest number of people to die, is the most studied disease due to its dangers, but it is difficult to thoroughly understand its characteristics and cause because of its complexity. A recent development of NGS technology has led to identifying the genomic features on a whole genome scale and cancer genome database such as The Cancer Genome Atlas (TCGA) has made clinicians and researchers use such genomic data to identify a clinical target, which can be applied for diagnosis and treatment. In order to find functional genes in cancer, RNA-sequencing is being carried out to analyze the changes in expression of genes between cancer and non-cancer samples. However, even within a single cancer type, not all patients have the same genomic and clinical characteristics. Therefore, knowing the differential transcriptomic profile in each patient group can improve the overall understanding of the different carcinogenic mechanism. In addition, genomic structural change such as gene fusion can bring variety to the expression of related genes which could be a potential mechanism for causing cancer. As a result,  comprehensive fusion gene profiling in cancer is also needed to understand the genetic complexity of cancer. To acquire such, we have obtained a large number of Korean hepatocellular carcinoma samples and produced massive transcriptome data. We are currently in the process of defining the transcriptome profiles from Korean hepatocellular carcinoma patients by subgrouping their molecular difference as well as tumor-specific fusion genes. By finding the differential targets among patient groups, we are attempting to develop a potential biomarker for treatment and prognosis of different hepatocellular carcinoma patients.

  With the advent of the NNGS represented by the likes of Oxford Nanopore Technologies, new vantage points to look far into cancer became available. Unlike previous short-read based technologies, longer read allows for improved assembly of each individual genomes, as well as detecting of complex structural variants (SV). Another clinical application which benefits greatly from the technology is that of transcriptome profiling. Each mRNA strand can now be sequenced without having to split into smaller pieces, where loss of information and inaccuracy are given. This new approach in RNA sequencing allows for much more accurate representation of transcriptome in cancer tissues as well as reduction in computation complexity, which in turn allows for much more confident novel findings. We are currently using this technology to look closely into Korean hepatocellular carcinoma patients and colorectal cancer patients, striving to find new insights into cancer by subtyping molecular differences, finding new cancer-specific isoforms, and developing new ways to further utilize the technology. Be it finding new prognostic biomarkers, or finding new drug targets, we hope to contribute to humanity's health. 



​Network Analysis of Cancer-related Long Noncoding RNAs

  In recent years, a breakthrough in RNA-seq technology has made it possible to see overall transcripts in cell. People found that more than 70% of the mammalian transcriptome consist of noncoding RNA (DJEBALI, Sarah, et al., Nature 2012). Long noncoding RNAs (lncRNAs) are a major class of noncoding RNAs which have 200 or more nucleotides in length. According to recent studies, lncRNA could regulate gene expression by various ways, including miRNA interaction, chromatin remodeling, transcriptional and post transcriptional regulation (BHAT, Shakil Ahmad, et al., Non-coding RNA Research 2016). And many paper report that dysregulated lncRNAs have been associated with many diseases such as cancer. The importance of lncRNAs clearly exists but comparing to protein-coding genes, still we do not know the function of many dysregulated lncRNAs. Because understanding dysregulated lncRNAs and mRNAs is important for cancer patient diagnosis, the aim of our study is to determine which lncRNAs are differentially expressed in cancer patients and reveal the relationship between lncRNAs and mRNAs using weighted correlation network analysis.

​

​Estimating Tumor Microenvironment by Transcriptome-based Deconvolution Algorithm

  Analysis of inter- or intra-tumoral heterogeneity has been tackled through new technologies such as bulk RNA-seq deconvolution algorithm and single-cell RNA-seq. In these ways, transcriptome analysis provides an opportunity to discover new biomarkers and therapeutic strategies allowing the exploration of cancer genes and mutations, cellular compositions, or interactions between these cells (Cieslik et al., Nat. Genet. 2018). Upon the formation of tumors, they form tumor microenvironment (TME) by modifying their interactions with surrounding mesenchymal cells to support their survival and progression. TME contributes to inter- or intra-heterogeneity of cancers, which hampers the treatments trials. Especially, immune cells are involved in progression and/or suppression of tumor growth and, therefore, understanding of immune infiltration would shed lights on treatment approaches of cancers (Junttila et al., Nature. 2013). Cancer stem cells (CSCs) that possess the ability of self-renewal and differentiations like normal stem cells constitute small subpopulations of tumor tissue that possess the ability to generate entire tumors. Thus, they are shown to be a cause of recurrence and metastasis and contribute cancer complexity and heterogeneity (Kreso et al., Cell Stem Cell. 2014). These characters of cancers are directly associated with patients’ survival. Therefore, we are analyzing populations and proportions of cells by deconvolution method using bulk tissue RNA-seq and single cell RNA-seq data. Furthermore, we are searching for clinically associated features of TME in order to discover diagnostic and prognostic markers of cancers. 



​Cancer Methylome Analysis

​  DNA methylation of cytosine (especially the fifth carbon of cytosine) at CpG dinucleotides is a crucial epigenetic alteration associated with human diseases, including cancers. Regulation of these processes are mediated by the TET2 dioxygenase and DNA methyltransferases such as DNMT1, DNMT3A and DNMT3B. Methylation affects chromatin stability and gene expression by epigenetic modification without changing DNA sequences. Therefore, methylation alteration plays key regulatory roles in the epigenetic machinery. In addition, aberrant DNA methylation has been detected at an early stage of carcinogenesis and gradually increases with cancer progression (Stirzaker et al. Trends Genet. 2014). Typically, global hypo-methylation and CpG island hyper-methylation are hallmarks of many cancer types, regulating transcriptional silence of tumor suppressor genes. Consequently, revealing a cancer specific methylation pattern is considered as a critical biomarker for cancer detection. Compared to genetic and expression bio-marker, epigenetic marker exhibits several advantages for cancer detection. Similar abnormal methylation patterns are mostly cancer specific. Additionally, analysis of methylation has great stability and tolerance of heterogeneity of the samples. We aim to discover cancer diagnostic and prognostic marker using DNA methylation.