This project uses data from the 2016 study by Bradford, James et al “Whole transcriptome profiling of patient-derived xenograft models as a tool to identify both tumor and stromal specific biomarkers”. Patient-derived xenografts (PDX) are models of cancer in which human tumor samples are studied in a mouse environment. In this study they distinguished tumor and stromal compartments by associating the stromal compartment with the murine (mouse) transcriptomic profile.
This is an exciting project for those interested in translational research and biomarker use in cancer subtype characterization. Students will also gain experience implementing computational tools to process data, eliminate technical issues and use machine learning algorithms to extract meaningful insights from molecular datasets.
Key Concepts
Tumor and its Microenvironment (stroma) – extracellular matrix, immune system and tumor cells: In the tumor microenvironment, cancer cells directly interact with both the immune system and the stroma. It is firmly established that the immune system, historically believed to be a major part of the body’s defence against tumor progression, can be reprogrammed by tumor cells to be ineffective, inactivated, or even acquire tumor promoting phenotypes. Likewise, stromal cells and extracellular matrix can also have pro- and anti-tumor properties.
Breast cancer subtypes: Cancer is often classified by the organ inflicted (breast, lung, lymphatic system), however, further subtypes exist based on the Pam50 classification of the tumor, and in the case of breast cancer, hormone receptor status is another factor which can differentiate sub-types. Pam50 is a tumor profiling test that assigns a sub-type to tumors based on their genetic profile. The hormone receptor status indicates whether the estrogen and progesterone hormone receptor proteins are detected in the tumor sample. When these receptors are present, the hormones may drive the cancer’s growth.
Patient-derived Xenograft Models: PDX models are cancer models in which tumor cells from the patient are implanted into a model mouse. This resembles tumor natural environment and allows to study both tumor and environment (stroma) progression and development.
Practical Skills
RNA-Seq: RNA-Seq is a method for studying the transcriptome by using Next Generation Sequencing to measure the mRNA present in a cell. Next Generation Sequencing data has specific characteristics requiring computational methods to process and interpret the raw data, turning it into a useful resource for further analysis and interpretation. The short reads produced by the sequencing machine need to be aligned to a reference genome and their abundance accurately measured to estimate gene expression.
Generating a Table of Gene Expression: For this purpose, the RNA-Seq by Expectation Maximization (RSEM) method is implemented, which takes the mapped read data and uses a statistical method that converts the “count” of the reads into a “level of expression” number for the whole gene or isoform.
Classification and Feature Selection: Using the gene expression table we will apply several algorithms to classify sub-types of samples and extract a small number of biological features that are most informative for distinguishing the samples into subtypes. This ability to identify sub-groups of samples based on their gene expression, and identify a small number of the genes that can be used to successfully predict the type of cancer allows us later to perform more specific studies focused on these features and the pathways they are involved in.
The full dataset contains 79 PDX samples, including 37 lung and 19 breast cancer samples (19 x breast, 37 x lung, 8 x colorectal, 7 x ovarian, 3 x endometrial, 2 x pancreatic, 2 x ampullary, 1 x leukemia): http://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-3980/.
