The development of whole transcriptome sequencing technology in the bulk-tissue and single-cell level has enabled the discovery of novel disease mechanisms in cancer and immunity research. At the cross-road of data science and biology, we adopted a data-oriented approach and developed methods that untangle RNA-seq data to address a variety of biological questions.

In the first study, we designed the Fast-Long Noncoding RNA Analysis (FLORA) pipeline for comprehensive study of long noncoding RNAs (lncRNAs) and applied FLORA to gastric cancer, leading to identification of thousands of GC-specific lncRNAs. Three lncRNA-based GC subtypes, i.e., L1, L2 and L3, were then revealed. Importantly, we demonstrated that the L3 subtype was an independent prognostic factor that robustly predicts pati...[ Read more ]

The development of whole transcriptome sequencing technology in the bulk-tissue and single-cell level has enabled the discovery of novel disease mechanisms in cancer and immunity research. At the cross-road of data science and biology, we adopted a data-oriented approach and developed methods that untangle RNA-seq data to address a variety of biological questions.

In the first study, we designed the Fast-Long Noncoding RNA Analysis (FLORA) pipeline for comprehensive study of long noncoding RNAs (lncRNAs) and applied FLORA to gastric cancer, leading to identification of thousands of GC-specific lncRNAs. Three lncRNA-based GC subtypes, i.e., L1, L2 and L3, were then revealed. Importantly, we demonstrated that the L3 subtype was an independent prognostic factor that robustly predicts patient survival outcome. Moreover, FLORA prioritized LINC01614 as a prognostic marker of GC, and subsequent perturbation experiments validated its functions in promoting GC proliferation and migration.

In the second and third studies, we profiled the mechanisms of noncoding RNAs in regulating B cell class-switch recombination (CSR). Firstly, we revealed an evolutionary conserved CTCF-mediated chromatin interactions in regulating lncRNA-CSR, which recruits protein complex and modulate CTCF occupancy. Secondly, we profiled the DIS3-knockout B cells and revealed genome-wide elevation of noncoding RNAs, decrease of CTCF binding, reduced chromatin interactions between Igh and the 3’ regulatory region, which contributes to CSR defect in B cell development.

In the last part, we revealed immune microenvironment in gliomas. Via deconvolution of bulk-tumor transcriptome data in secondary glioblastoma, tumor-associated macrophages were found and validated to be enriched in patients with MET alterations. We further revealed the single-cell immune landscape of glioblastoma and identified the tumor-associated monocytes were associated with EGFR signaling and angiogenesis.

Collectively, we showed that utilizing data science technologies to trace previously overlooked transcriptional signals has provided a novel angle to understand complex human disorders, including cancer and immunodeficiency.