CS 7960 Topics in Bioinformatics, Spring 2008

Homework 2 (100 points)

Due Time: 11:59pm, Feb 18 (Monday), 2008

In this assignment, we will use posterior decoding to solve the problem introduced in assignment 1 (i.e. to predict the topology of Alpha helical transmembrane proteins). Different from a assignment 1, here, we will use posterior decoding instead of viterbi decoding to find the topology of the proteins.

Please submit the following files through eagle:

1) Source Codes

2) A README file with instructions on how to compile and run your program

3) The predicted topology of each protein using posterior decoding.

4) A simple report of the performance of the method. Are there any invalid transitions? If so, how many of them are invalid? Why?

Notice: In assignment 1, we used the sum of log odds to solve the underflow problems. In this assignment, we will use forward and backward algorithm to calculate the posterior possibility. The forward/backwark algorithms sum probability values, so it is not a viable solution to log the values in order to avoid underflow. Here, will use scaling coefficients to avoid underflow. Please find the details in section 6 (Implementing HMMs) of this paper.