2055: Biost Homework 1
Distributed: 1/16/09 Due: 1/26/09
1. (5 points) a) p16 is a known tumor suppressor gene involved in cell cycle regulation. Use NCBI database to find its
official gene symbol as CDKN2A. What other aliases this gene has? What is the Entrez ID of this gene? (print the page where you find the information).
b)
Use Map Viewer to retrieve sequence information of CDKN2A gene (with 2Kbp upstream and downstream regions) in Homo sapiens (human) and save it as a text file named CDKN2A.txt. How long (how many base pairs) is this sequence?
c)
Use Bioconductor to similarly retrieve sequence of CDKN2A with 2Kbp upstream and downstream. How long is this sequence? (There should be two alternative splicing forms of this gene. Pick the one identical to that from NCBI database.)
2. (5 points) It is known that the dinucleotide combination CG is often enriched in upstream or downstream
region of a gene (so-called CpG island). The area can be methylated to silence the expression of the gene. a) Generate histogram plot of CDKN2A sequence obtained from 1(c) and show that CpG is slightly enriched in the upstream region. (Note that CDKN2A is transcribed from the negative strand in opposite direction from NCBI Map Viewer.)
b)
Perform hypothesis testing to test whether CDKN2A has enriched CpG islands in 2K upstream region. Establish a 22 table similar to the BRCA1 example. Perform Pearsons chi-square test to show that the enrichment is statistically significant.
c)
An alternative test is to randomly simulate gene sequences to generate the null distribution of the CpG count. (1) Generate a random sequence of the same length of CDKN2A with the same sampling probability of ATGC in CDKN2A (2) Compute the number of CpG appearance in the 2K upstream region in the simulated sequence Repeat (3) step (1)-(2) for B=1,000 times to obtain a null distribution of the CpG occurrence count (4) Compare the observed CpG occurrence count to the null distribution to assess p-value.
(optional bonus 2 points) d) A third alternative test is to randomly retrieve the sequence of a gene by Entrez ID. Since each gene have different background CpG occurrences and different sequence length, a more reasonable statistic is to use the odds ratio of CpG occurrence in 2K upstream region and in the coding sequence. Use this idea to construct a test to assess the p-value of CDKN2A. Note: From 2B-2D, we see how three tests might be applied to a same biological question by different statisticians and how we decide which is better. Usually two key factors, computation feasibility and fitness of assumption, should be considered. The chi-square test in 2B is computatioinally fast while the independence assumption is violated. The test in 2C is a little better but its slow and the independence assumption in sequence simulation is not true in general. The test in 2D assumes that ma...