Project 1 notes

0. Use the new and improved pickCA.m to read multiple chains ('help pickCA').

1. Use interncoord.m to compute internal coordinates. This code is part of the solution to HW1.

2. For secondary structure annotation, you can either refer to SHEET and HELIX lines in the preamble of 1MHC.pdb, or you can go to the 1MHC page at www.pdb.org and follow the links Geometry->Motif Summary.

3. The bond angle ambiguity referred to in class was a false alarm. Although a bond angle can only be between 0 and pi, the corresponding torsion angle determines whether the rotation is "to the left" or "to the right."

4. For the secondary structure annotation, feel free to drop a couple of amino acids at the ends of the chains, in order to match up the number of amino acids with the numbers of bond lengths, bond angles, and torsions. Explain what you do here clearly and briefly.

5. Submissions are due at the beginning of lecture this Thursday, October 9. Hardcopy submissions are due at the beginning of class. Submit electronically as follows (similar to HW1). Place all files in one directory named after your CU NetID. Compress this directory into one file using zip or gzip (no other applications). Email this file to tony@cs.cornell.edu as an attachment with nothing else in the body and with the subject CS321_2003fa_P1 (please observe case and include no spaces).

6. Submit the following output in addition to your code. Remember that your code and output should be presented in the order of the questions. Also remember to label the axes on all plot or graphs and to give them titles.

• 3rd bullet: One plot showing all six chains in six different colors. This is done by the function plot_chains.m discussed in class.
• 4th bullet: Compute only distances between consecutive residues, not the whole distance matrix. For the histogram, lump the data from all six chains into just one histogram. See the solutions to HW1 for how to build a histogram. Feel free to (re)use here the function interncoord_m.m described below.
• 5th bullet: One plot analogous to the plot from the 3rd bullet. This is done by the function prot_spline.m discussed in class. It can call plot_chains.m after it's done interpolating.
• 6th bullet: Show only the function that calls interncoord.m on each of the 6 chains. You can call this function interncoord_m.m. It should work for a protein of m chains, where m can be anything (don't hardcode the number 6). interncoord_m should return bond lengths, bond angles, and torsion angles, e.g., in three cell-arrays of length m.
• 7th bullet: No output other than the code.
• 8th bullet: Three correlation plots: one for bond length, one for bond angle, and one for torsion angle.
• 9th bullet: No output other than the code.
• 10th bullet: Show only one pair of contact maps, for the A chain of MHC before and after reconstruction. For extra credit, compute the difference between the two distance matrices (used to make the contact maps) as the maximum absolute value of element-wise differences (show code and output). If your code is correct, this number should be smaller than 1e-9.

7. Remember that brevity is a virtue. To give you an idea of how simply the code can be written, here are the sizes of some of the functions used in the solutions. These are the numbers of lines in function bodies, not including function comments or blank lines:

• plot_chains.m: 13 lines
• prot_splines.m: 5 lines
• alignmat.m: 23 lines
• rotmat.m: 7 lines (calls alignmat.m)
• intern2cart: 20 lines (calls rotmat.m). This is the function that constructs a protein's Cartesian representation from internal coordinates: r = intern2cart(d, a, t).

8. For those of you who don't have dist.m and normr.m etc., these functions are now linked to the matlab code page, from the main course web page.

9. Enjoy.