Frequent Mutations in Cancer

Read the paper Vogelstein et al., Cancer Genome Landscapes, Science, 2013 http://www.sciencemag.org/content/339/6127/1546.full Fifteen years ago, the idea that all of the genes altered in cancer could be identified at base-pair resolution would have seemed like science fiction. Today, such genome-wide analysis, through sequencing of the exome or of the whole genome, is routine. (Vogelstein et al., 2013)

The paper presents a comprehensive study about mutations status and their effect in cancer. They point out that there is a fundamental difference between a driver gene and a driver gene mutation. A driver gene is one that contains driver gene mutations, but may also contain passenger gene mutations which do not drive the cancer. The best way to identify Mut-driver genes is through their pattern of mutation rather than through their mutation frequency. The patterns of mutations in well-studied oncogenes and tumor suppressor genes are highly characteristic and nonrandom. Oncogenes are recurrently mutated at the same amino acid positions, whereas tumor suppressor genes are mutated through protein-truncating alterations throughout their length. (Vogelstein et al., 2013)

The authors propose a method, the 20/20 rule, which classifies genes in oncogenes and tumor suppressor genes based on the type and the frequency of mutations. Using data from Catalogue of Somatic Mutations in Cancer (COSMIC) they computed a score that is able to predict whether a gene is more likely to behave as an oncogene or a tumor suppressor gene. The paper provides a list of 125 genes and their oncogene and tumor suppressor gene scores.

Note

Roles have not been specified in this project. Work with your team to divide up the tasks and describe how you did this in the Methods section of your report.

Upon completion of this assignment, students will be able to do the following:

  • parse, subset, and tabulate a large tab separated table containing mutation data
  • compute a score based on genomic coordinates to prioritize oncogenes and tumor suppressors

1. Read the paper

Vogelstein, Bert, Nickolas Papadopoulos, Victor E. Velculescu, Shibin Zhou, Luis A. Diaz Jr, and Kenneth W. Kinzler. 2013. “Cancer Genome Landscapes.” Science 339 (6127): 1546–58. PMID: 23539594

2. Examine the COSMIC Targeted Screen Mutation Dataset

The COSMIC mutation database is the largest and most comprehensive collection of cancer mutations from human tumors available. There are two types of datasets: targeted screen and genome-wide. The targeted screen mutations have been carefully curated by experts to include only high-confidence mutations, while the genome-wide mutations were created by analysis of large sets of pubilcly available tumor genome data.

The Targeted Screening dataset has been made available for you at /project/bf528/project_4_snps/data/CosmicCompleteTargetedScreensMutantExport.tsv and can be used for all parts of this assignment.

  1. Familiarize yourself with the data format for the “COSMIC Complete Mutation Data (Targeted Screens)” dataset in the COSMIC database downloads page. The file is tab separated with 35 columns.

  2. Compute the following statistics using the Targeted Screen mutation dataset and put them in a table:

    • Total number of samples
    • Total number of mutated samples
    • Total number of negative samples
    • Total number of tumors
    • Total number of genes
    • Total number of mutations

  3. Produce bar plots or pie charts for each of the following distributions:

    • Number of mutations broken out by mutation description
    • Number of mutations broken out by Primary site
    • Number of mutations broken out by Primary histology

    Some of the values for these fields have very low counts. You might consider grouping low-count values into an ‘Other’ category for readability.

Deliverables:

  • a table with the statistics from 2.2
  • three bar or pie charts showing the distribution of mutations according to primary site and histology

3. Detailed analysis of mutations per tumor by cancer type

  1. Produce a matrix that records the number of tumors broken out by Primary site vs Primary histology. For example, there are 13,390 mutations in breast tumors of type carcinoma. Visualize this matrix as a heatmap with appropriate labels.
  2. Identify the number of mutations in each tumor grouped by cancer type, as in Figure 1B. Produce a boxplot of tumor mutation distributions with the x-axis ordered like that of the figure.

Deliverables:

  • a heatmap showing the distribution of site vs histology type
  • a boxplot of the distribution of number of mutations per tumor similar to Figure 1B

4. Mut-driver genes and the 20/20 rule

Some genes are more likely to be mutated in certain cancers than others. However, the authors note that a simple overabundance of mutations in a gene is not sufficient to explain the tumorigenicity observed in the sample.

  1. Using the mutated and negative samples, calculate the fraction of mutated vs. all tumors for each gene in the dataset. Examine the top genes when sorted by absolute number of mutated tumors compared with fraction. Report the top ten genes sorted by both of these metrics in tables.
  2. One potential factor in the calculation of mutation frequency is the length of each gene. Use the Gene CDS Length column to find the coding sequence length and plot it against the number of mutations calculated in 4.1 as a scatter plot. Note any trends you observe in the plot.
  3. Implement the 20/20 rule as described in the paper. Use the Mutation genome position column and the description of “Missense” for mutated tumors to identify recurrently mutated positions for oncogenes. Use mutations with description “Nonsense” to identify mutations that are inactivating for tumor suppressor genes. Include the top ten oncogenes and top ten tumor suppressor genes as ranked by highest 20/20 score in separate tables.
  4. Compare your lists of top scoring genes to the 125 Mut-driver genes listed in the manuscript supplemental table S2A.

Deliverables:

  • tables for the top ten genes sorted by number of mutations and fraction mutated
  • a scatter plot of fraction mutated versus CDS length
  • tables for the top ten oncogenes and tumor suppressor genes ranked by 20/20 score
  • a comparison of your 20/20 scores to those reported in the paper

5. Mutation frequency by tumor and cancer type

For 4 person groups only

  1. Extend the analysis from part 4 by subdividing the fraction of mutated vs all tumors by Primary site and Primary histology. Choose any three of each type and identify the most frequently mutated genes for those types. Report the top ten most frequently mutated genes for each type in a table with the frequency.
  2. Use the top 100 genes from each of your chosen types from 5.1 and run a gene set enrichment analysis using the tool of your choosing. Report any significantly enriched gene sets in your report.

Deliverables:

  • a table containing the top ten most frequently mutated genes for three sites and three histology types along with their frequency
  • any significant gene set enrichment results for the top 100 genes ranked by mutation frequency

6. Discuss Your Findings

Discuss your findings with your team members and other teams who chose this project. Some interesting questions to consider:

  1. Is the distribution of sites and histology types in this dataset consistent with the relative frequency of these cancer types as reported elsewhere in the literature? Why or why not?
  2. Is the frequency of site vs histology type consistent with your expectations? Are there any site/histology type combinations that are unexpected?
  3. Why isn’t mutational frequency alone a good predictor of tumorigenicity?
  4. Do your overall results agree with those presented in the paper?

Assignment Writeup

Project Writeup Instructions