This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: CS228 Programming Assignment #2 1 Stanford CS 228, Winter 2011-2012 Assignment #2: Bayes Nets for Genetic Inheritance This assignment is due at 12 noon on January 23 . 1 Overview Because of your success in modeling credit-worthiness, your fame as an expert in graphical models has spread, and now genetic counselors are seeking your assistance. Genetic counselors want your help in advising couples with a family history of a genetic disease. Speci cally, they want to help such couples decide whether to have a biological child or to adopt by assessing the probability that their un-born child will have the disease. Through this assignment, you will see how Bayesian networks can be used to determine such probabilities through modeling the mechanism of genetic inheritance. During this assignment, you will use genetic information to predict the probability that a person will have a physical trait. Many human physical traits, such as freckles, hair color, and many diseases 1 are controlled by one or more proteins. These proteins are coded for by genes, which are sequences of DNA that are found in every cell and passed down from generation to generation. Genetic inheritance patterns 2 are generally consistent from generation to generation, so template models are a natural way to model them. If you do not know much about genetics, you may nd the Khan Academy Introduction to Heredity Video helpful http://www. khanacademy.org/video/introduction-to-heredity?playlist=Biology . There is also an on-line Appendix, and you will be directed to it during the assignment so that you have the appropriate background to complete the di erent sections. Genetic counselors will be giving you pedigrees (family trees), allele frequencies for di erent alleles, and some information on the e ects of having di erent alleles. You will construct Bayesian networks from this information and determine the probabilities of having di erent genetic traits. You will then use your network to answer questions that a genetic counselor might encounter. The questions can be found in PA2 Quiz (22 points, 2 points for each question) . Note that, if you are working with a partner, both you and your partner need to submit the quiz and the programming assignment separately. You are allowed to submit the quiz twice, and you may submit your code as many times as you want (but only the last submission will be graded). This assignment will cover some facets of Bayesian networks that were not covered in PA1. You will implement a complete Bayesian network , programming the entire thing from scratch! You will experience how using templates allows you to copy CPDs. You will also be exposed to some of modeling choices in constructing Bayesian networks and see how adding nodes sometimes allows you to dramatically simplify CPDs. In addition, you will get to work with both table and sigmoid CPDs. Note that, while the description of this assignment may seem long, little coding is required ....
View Full Document
- Winter '09