Genetics 144, Oncogenomics

Darmouth Medical School

Winter 2005, weekly notes and discussion topics

Charles Brenner, Ph.D.

charles.brenner@dartmouth.edu

For the latest update, please reload http://www.dartmouth.edu/~brenner/notes.html


Notes for week 1, chapter 1:
At the Precarious Cusp of Oncogenomics

1-1. How to read Oncogenomics: situate yourself in a quiet place near a high speed net connection, your course notebook, and your favorite cell and molecular cell biology reference book.
1-2. Please collect typos and errors and bring them to my attention. Also, please check URLs and suggest additional online resources.
1-3. It would surprise me if you can read many chapters without looking up some of the underlying literature. Some of the underlying literature will interest you enough to want to delve more deeply, potentially into more recent papers.
1-4. Everyone will be responsible for an oral and web-based presentation. Make sure that your web presentation can be understood without your oral narrative. It is unlikely to be exactly the same document you present in class. If you have to miss a class, please write up thoughtful answers to one or more discussion questions and email them to me.
1-5. Cancers are many different diseases, traced to many different genotypes.
1-6. Cancers are diseases of escape.
1-7. Four types of genetic changes occur in the development of cancer.
1-8. Describe the natural selection that occurs at the cell level in neoplastic development.
1-9. Are most cancers sporadic or hereditary? Is there a sporadic component to hereditary cancer? Is there a hereditary component to sporadic cancer?
1-10. What is an oncogene? What kind of mutations occur in oncogenes? Are these somatic alterations or germline alterations?
1-11. What is a tumor suppressor gene? How can an allele of tumor suppressor gene mutation appear to dominant at the individual level and be recessive at the cell level?
1-12. What is penetrance? Define "age of onset," "age of diagnosis," multifocal, bilateral.
1-13. Why would low penetrance or polygenic traits make it hard to see hereditary cancer syndromes?
1-14. What is the difference between a genotype and a gene expression pattern?
1-15. What is epigenetics?
1-16. What makes something a genetically validated drug target? What makes a target "druggable"?
1-17. What are modifier genes and what might they have to do with drug targets?
1-18. What is the double-edged sword of molecularly targeted clinical testing?

Notes for week 2, chapter 2:
Genome-wide Searches for Mutations in Human Cancers

2-1. Stratton points out that mapping chromosomal alterations has been the most successful method to identify cancer genes. Why isn't this a general solution to identifying all genes causally associated with cancer?
2-2. Is copy-number analysis more or less challenging with a tumor biopsy versus a cancer cell line? Why?
2-3. How does the assay depicted in Figure 2-1 work for 5500 different loci at once?
2-4. What are the arguments against mutational screening of cDNA?
2-5. In order to detect rearrangements by end-sequencing, what would you have to do?
2-6. If the CGP were to use a single reference DNA in the intragenic mutation detection system depicted in Figure 2-2, what kinds of differences would they not be able to distinguish?
2-7. What was suprising and not suprising about the discovery of BRAF mutations in melanoma?

Notes for week 3, chapters 3 and 4:
Array-based Comparative Genomic Hybridization and SNPs

3-1. In Box 3-1, why are the normal and tumor DNA samples called targets and the arrayed elements called probes? Note that there is an exact terminological answer to this question.
3-2. What are the arguments for and against classifying HOXB7 as an oncogene? What would you like to see done next with HOXB7?
3-3. From the selfish gene point of view, how might a low-penetrance cancer predisposition allele become more prevalent in the population than a high-penetrance cancer predisposition trait?
3-4. Remember the following basics of population genetics. For linkage, you need families and what you identify are chromosomal regions, initially much larger than a single gene. For association, you need populations and you have the potential to pinpoint the disease gene. However, the frequencies of many different alleles differ between populations and this must be taken into account in association studies. Thus, if you wanted to find SNPs associated with prostate cancer in African-Americans, you would be looking for SNPs preferentially found in the affected individuals versus not found in the same ethnic group. Otherwise, all you would find are SNPs associated with being African-American.
3-5. How does a haplotype and a high density of SNPs in a region help association studies?
3-6. What would be your criteria for establishing the bona fides of a disease-associated SNP or haplotype?

Notes for week 4, chapters 5 and 6:
RNA Expression Profiling and Cancer


4-1. Please realize that when all of the primary data in a microarray experiment are reduced to the log ratio of 1 target's hybridization to another target's hybridization (both to the same probe), all of the intensity information that would allow you to compare the expression of one probe to another is lost. What are the practical reasons why these data are discarded? How might you restore intensity data to a clustergram? What could you do with these data if you could reliably measure them?
4-2. What are some of the arguments in favor of a universal reference RNA?
4-3. Explain hierarchical clustering.
4-4. Explain supervised clustering.
4-5. What is the argument that, at least in some cases, an expression pattern is more powerful than a genotype?

Notes for week 5, chapters 7 and 8:
The NCI-60 and Tissue Analysis in Cancer


5-1. How does the Weinstein group define targets? Do they think there is a single target in a particular tumor? Are they only embracing oncoproteins as targets? What types of complexities are embraced by including gene expression patterns and genetic losses as targets?
5-2. Thinking about probes and targets, why is a tissue microarray more like a Northern or Western than an expression array?
5-3. What aspect of tissue microarrays or makes it possible to analyze multiple probes?
5-4. In a laser capture microdissection study or layered expression scanning study, how do you know anything about the evolution of a tumor if you don't have earlier or later timepoints?

Notes for week 6, chapters 9 and 10:
Proteomics and Nonmammalian Models


6-1. Is it conceivable to use a proteomic approach to classify tumors in a useful way (such as therapy responsiveness) without knowing what the specific proteins are?
6-2. If you had a strong correlation between a specific drug sensitivity and presence of a particular protein profile (or one protein/peptide species), how would you propose to determine the drug mechanism of action?
6-3. Provide some real and imagined examples of "genetically validated drug targets" for carcinogenesis, metastasis and maintenance of the malignant phenotype along with the type of experimental evidence you would like to see. Can you think of any examples in which genetic validation has been accomplished by "chemical genetics?"
6-4. A drug is relatively toxic to yeast cells and leads to a specifically altered gene expression profile. What genomic experiment could you do to try to identify the primary target? If you identify the primary target and it is a nonessential gene, what effect should the drug have on deletion mutants for that gene?

Notes for week 7, chapters 11 and 12:
Mouse Models for Analysis and Discovery of Cancer Genes


7-1. Many modern mouse models involve conditional inactivation or activation of cancer genes in a time and/or tissue-specific manner. As the number of cells that bear the preneoplastic genotype are reduced, the model is usually considered to be more similar to human cancers. Are there any downsides to the time, tissue and cell number restrictions that these models create? What are the solutions?
7-2. What are the arguments for performing reverse mouse genetic experiments in a mixed genetic background versus multiple pure backgrounds?
7-3. What type of human cancer genetics is most similar to QTL studies of modifier genes?

Notes for week 8, chapters 13 and 14:
Drugs Targeting Kinases and the Ras Pathway


8-1. In what was is anti-angiogenic pharmacology genotype-specific versus genotype-nonspecific? What are the advantages and disadvantage or genotype-nonspecificity?
8-2. Is there any advantage to a drug having multiple targets? What is meant by "therapeutic window"?
8-3. How do you explain (molecularly and cellularly) the addiction model of oncoprotein function? Why should oncoprotein inhibition cause death rather than cell cycle arrest?
8-4. If checkpoint kinases are required for normal cell functions, why are these molecules considered to be attractive cancer drug targets? How important do you think genotype and targeted agents will be in combination with radiation and conventional chemotherapies?
8-5. If you were to initiate a program for compounds that disrupt Ras signaling, what assays would you propose using for screening? What are the downsides to the assays that you would consider and not consider using?


Notes for week 9, chapters 15 and 16:
Genome-Informed Clinical Testing and the Future of Oncogenomics


9-1. Why have pharmaceutical companies shied away from genotype-informed trials?
9-2. What are the technological reasons that make genotype-informed trials difficult?
9-3. What are the endpoints that should be measured in a genotype-informed trial?
9-4. For which cancer types might you be comfortable doing a trial with a targeted drug without first determining genotypes?
9-5. How far along do you think we will be in 2010 and 2015?

Back to the Genetics 144 Home Page.

Back to the Brenner Group Home Page.