Li Lab for Cancer Genomics and Computational Cancer Immunology Research Overview

The human body is able to produce 10^16 types of different ⍺β T cells, with each T cell possessing a unique set of the hypervariable receptors. The collection of T cell clonotypes, referred to as the T cell repertoire, is extremely diverse and responsible for the recognition and elimination of a vast pool of external and internal pathogenic antigens, such as virus, bacteria and cancer antigens.

Our team is interested in learning the connection between individual's immune repertoire and the antigenic landscape. An ongoing research direction in the Li lab is to use bioinformatics and genomics approaches to study the interaction between tumor-infiltrating T cell repertoire and the malignant cells. We are developing a novel algorithm that predicts closely related CDR3 groups based on protein sequence structures, to study the immunogenomics data and understand the antigenic landscape of human perihperal T cell repertoire.

The Darwinian process during cancer evolution produces a heterogeneous pool of malignant cells with a distinct landscape of somatic changes in the genome. Cancer cells are under constant selective pressures, including immune cell attack, drug cytotoxicity, lack of nutrition, and more. The “winning” clone almost invariably overcomes environmental disadvantages and grows into an end-stage tumor.

The Li Lab is interested in understanding the process of co-evolution of tumor and immune cells during cancer development, which can be tracked from clonal expansion events, together with components of the tumor microenvironment and infiltrating immune repertoire. One frequent consequence of this co-evolution is the avoidance of almost all immune surveillance, including cytotoxicity mediated by CD8+ T cells or auto-antibodies. Our team will continue to investigate somatic changes involved in different immune evasive pathways to identify novel targets for cancer immunotherapies.

Immune checkpoint blockage (ICB) therapies have achieved remarkable clinical success in the treatment of multiple end-stage tumors, including melanoma, bladder cancer, Hodgkin lymphoma, non-small cell lung carcinoma, and kidney, head and neck cancers. However, accepting ICB treatment remains a difficult decision for patients with cancer and their families due to prohibitive costs, undesirable side effects, and low response rates. Therefore, it is critical to develop predictive biomarkers for ICB therapies to better inform clinical decision making. Our team collaborates with Dr. Yang-Xin Fu to study the immune contexture of the cancer microenvironment through work with animal models, with a focus on the identification of novel prognostic biomarkers to guide ICB therapies.

Researchers in the Li Lab developed a collection of innovative and widely used open access bioinformatics tools.

TRUST (Tcr/Bcr Receptor Utilities for Solid Tumors)

TRUST extracts T/B cell receptor hypervariable CDR3 sequences from unselected tumor RNA-seq data. It is an ultra-sensitive de novo assembly method for calling CDR3s, with demonstrated utilities when applied to large cancer genomics data. It is written in Python2, with continued updates for performance improvements and additional functionalities. Currently, TRUST supports multiple RNA-seq aligners including Bowtie2, STAR, MapSplice2, and more. It also applies to both hg19 and hg38 human genome references. TRUST can run in parallel mode and uses GPU acceleration to increase computational efficiency.

TIMER (Tumor IMmune Estimation Resource)

TIMER originally referred to the statistical deconvolution method developed to estimate six different immune cell types in the tumor microenvironment using gene expression data. Tumor purity estimation for more than 9,000 TCGA tumors were kept as a resource. The team developed an interactive website where users can perform a number of analysis using the estimated immune infiltration levels we estimated.

CHAT (Clonal Heterogeneity Analysis Tool)

CHAT is a collection of tools for studying intra-tumor heterogeneity using genomics data, borne out of the team's development of a tumor purity inference algorithm and revisiting glioblastoma multiforme classification. Modifications were made to the framework of this algorithm to allow the inference of the subclonality fractions of tumor somatic copy number alterations (sCNA). Integrative analysis of sCNA data with whole exome sequencing data enabled the estimation of cancer cell fraction of the somatic mutations. The temporal order for a pair of somatic events was also estimated when available.

iSMART

iSMART (immuno-Similarity Measurement by Aligning Receptors of T cells) performs a specially parameterized pairwise local alignment on T cell receptor CDR3 sequences to group them into antigen-specific clusters. iSMART is under GPL3.0 license, written in Python2, and portable without installations.

DeepCAT

DeepCAT (Deep CNN Model for Cancer Associated TCRs) is a computational method based on a convolutional neural network to exclusively identify cancer-associated beta chain TCR hypervariable CDR3 sequences. The input data were generated from tumor RNA-seq data and TCR repertoire sequencing data of healthy donors. Users do not need training or evaluation to run the cancer score prediction function in DeepCAT.

GIANA

GIANA (Geometric Isometry-based TCR AligNment Algorithm) is a computationally efficient tool that provides the same level of clustering specificity as TCRdist at 600 times its speed without sacrificing accuracy. GIANA allows the rapid query of large reference cohorts within minutes and the ability to cluster large-scale TCR datasets, providing candidate disease-specific receptors and a new solution for repertoire classification.

MERCI

MERCI (Mitochondrial-enabled Reconstruction of Cellular Interactions) is a statistical deconvolution method for tracing and quantifying mitochondrial trafficking between cancer and T cells. Through rigorous benchmarking and validation, MERCI accurately predicts the recipient cells and their relative mitochondrial compositions. Application of MERCI to human cancer samples identifies a reproducible mitochondria (MT) transfer phenotype, with its signature genes involved in cytoskeleton remodeling, energy production, and TNF-α signaling pathways. Moreover, MT transfer is associated with increased cell cycle activity and poor clinical outcome across different cancer types.