Powered By Bing

Yu Lab: Research

Research Overview

Biological processes operate through molecular networks at the cellular level and cell–cell networks at the tissue/organ level. Deciphering the “wiring” and “rewiring” of these networks under normal and pathologic conditions is a fundamental, yet challenging goal of biomedical research. The research of the Yu laboratory is focused on developing data-driven systems biology algorithms to integrate omics data, collected in bulk or at the single-cell level, to decipher these “wiring” and “rewiring” events, as well as “hidden” drivers underpinning the biological processes in health and diseases, and to translate the in-silico discoveries into therapeutic targets, biomarkers, and combination therapies for human cancer and other disorders. We have been collaborating with cancer biologists, immunologists, and clinicians to complete these studies.

Hidden Driver Inference

Signaling proteins and epigenetic factors are crucial drivers of network rewiring and are likely druggable, making them ideal therapeutic targets. However, it is often difficult to unbiasedly identify many of these drivers (referred to here as hidden drivers) because they may not be genetically altered or differentially expressed at the mRNA or protein levels but, rather, are altered by posttranslational modifications (e.g., phosphorylation) or other mechanisms. In collaboration with Dr. Hongbo Chi, we have developed NetBID (Du et al., Nature 2018), a data-driven network-based algorithm to infer hidden drivers by integrating multi-omics data, which successfully identified Hippo kinases as hidden drivers of CD8α+ dendritic cells for antitumor immunity. We have also developed, SJARACNe (Khatamian et al., Bioinformatics 2018), a scalable version of the ARACNe algorithm for reverse-engineering both transcription regulatory and signaling networks from large-scaled gene expression profiles at both bulk and single-cell levels. NetBID has been increasingly used in many contexts, for example, identification of hidden drivers of glucocorticoid resistance in pediatric leukemia (in collaboration with Dr. Bill Evans; Autry et al., Nature Medicine, in revision).

NetBID model

Translational Systems Biology

We have been working closely with cancer biologists and oncology clinicians to translate our systems biology discoveries of hidden drivers and networks into potential therapeutics and biomarkers for cancer treatment. For example, we have successfully translated two of our discoveries into two clinical trials for breast cancer, in collaboration with Dr. Jose Silva at Mount Sinai and Drs. Andrea Califano and Kevin Kalinsky at Columbia University. One trial is targeting the JAK/STAT3 pathway with ruxolitinib and the HER2 pathway with trastuzumab in metastatic HER2+ breast cancer (clinical trial: NCT02066532), where our systems biology analysis predicted that STAT3 is a hidden driver in HER2+ breast cancer, specifically in the ER−/HER2+ subpopulation (Rodriguez-Barrueco & Yu et al., Genes & Development 2015). The clinical results so far have confirmed the specific efficacy of combining ruxolitinib with trastuzumab in patients with ER−/HER2+ breast cancer. The other trial examines an HDAC6 inhibitor (ACY-1215) in combination with chemotherapy (paclitaxel) for metastatic breast cancer (clinical trial: NCT02632071) and originated from our discovery of HDAC6 as a non-oncogene addiction driver for inflammatory breast cancer (Putcha & Yu et al., Breast Cancer Research 2015). According to the clinical results obtained thus far, our HDAC6 score based on a data-driven HDAC6 subnetwork predicts patient response better than the conventional ER+/HER2− status, which is also part of our predictions. Our unbiased discovery of AKT as a driver of glucocorticoid resistance in T-cell acute lymphoblastic leukemia suggested the potential application of AKT inhibitors to overcome the resistance in the clinic (Piovan & Yu et al., Cancer Cell 2013).

HDAC6 model

At St. Jude, we are developing network activity–based biomarkers for novel targeted therapies for pediatric leukemia (in collaboration with Drs. Jun J. Yang and Tanja Gruber). We are also developing data-driven network-based algorithms to predict synergistic drug combinations and perform targeted drug combination screens to identify novel combination therapies for pediatric brain tumors (in collaboration with Drs. Martine F. Roussel, Suzy Baker, and Amar Gajjar).

Single-cell Systems Biology

Based on our expertise in bulk data–based systems biology, we are developing computational algorithms for network inference and analysis from single-cell RNA-seq (scRNA-seq) data. Specifically, we develop mutual information–based algorithms for clustering analysis, cell type–specific inference of intracellular networks, hidden drivers and network rewiring, and inference of intercellular communication networks. We also develop algorithms to integrate single-cell results with bulk-analyzed patient data for translational insights into biomarkers and therapeutics.

Single-Cell Systems Biology

Systems Immunology and Immuno-Oncology

We work very closely with Dr. Hongbo Chi from the Department of Immunology and Dr. Junmin Peng, Director of St. Jude’s Center for Proteomics and Metabolomics, to develop and apply systems biology algorithms to answer fundamental questions in immunology by integrating multi-omics data (Yu, Peng, and Chi. Current Opinion in Systems Biology, in review). For example, we have identified Hippo kinases as hidden drivers of conventional dendritic cell subset 1 for T-cell priming and antitumor immunity by NetBID algorithm (Du et al., Nature 2018). We use the latest technologies of bulk RNA-seq, scRNA-seq, scATAC-seq, and proteomics to profile immune cells in healthy and pathologic (e.g., cancer, inflammation, Alzheimer’s disease) contexts. We then use systems biology approaches to integrate those omics data and dissect the multiscale networks and to identify hidden drivers and network rewiring of various immune cell types (with a focus on T cells, including T regulatory cells, CD4/8+ T effector cells, and T helper cells) in homeostasis, development, and dysregulation of the immune system.

We use systems biology approaches to improve immunotherapy for cancer treatment by working with Drs. Stephen Gottschalk and Chris DeRenzo to identify biomarkers of CAR-T cell therapy and key determinants/modulators to improve CAR-T persistence. We also work on new indications of immune checkpoint blockade therapies and potential combination strategies for pediatric solid and brain tumors.

Systems Immunology

Functional Genomics (RNAi/CRISPR screens)

We have extensively used functional genomic CRISPR screens to complement our systems biology analysis to identify hidden drivers and therapeutic targets. We also have developed the ScreenBEAM algorithm (Yu et al., Bioinformatics 2016) by using Bayesian hierarchical modeling for gene-level scoring from noisy RNAi/CRISPR screening data. A new version of ScreenBEAM that overcomes the imbalance problem of library size and has more features is under development.