Monday, 30 June 2014

week 3 but a bit late!

Well I am beginning to feel more confident in python and decided to implement the alignment of two structures as was done by W. Kabsch (1976). Nice bit of maths and also a fun introduction to eigenvectors! Though wasted two days on learning that np.eig wont work on symmetric matrices and one should use .eigh,.

Well at least I now have my alignments for the ILE rotamers  and calculated all my RSMDs to compare the geometry optimized structures against the ones I produced earlier.

So I found the alignment looked best (example shown below made using Pymol) when the alignment was done using only the fixed atoms to generate the rotation matrix.

Friday, 20 June 2014

Week 2 Many plots!

This week I have managed to make a Ramachandran plot for ILE using the top500 pdb files. It seems like the majority of the ILE phi psi angles fall in the beta-sheet and alpha-helix regions with very few in the left hand alpha-helix region. This was expected as the left hand alpha-helix is much rarer.

Following this I also made plots of the chi 1 and chi 2 angles according to their secondary structure. This gave some interesting results showing that in alpha helix the  x1 tended be in distinct regions. There was also some interesting results as when the back bone was in the beta sheet conformation there was a population in the chi 2 in +100.

This is just one of the 4 plots! also made the for ILE residers depending on the backbone conformation.

(disclaimer the figure is not finished yet!)

Based on these plots I have taken a selection of angle combination for phi, psi, chi 1 and chi2 and generated the structures. I hope to be able to optimise them soon but it seems I still have trouble with writing the input files for Gaussian 09. Oh well at least I understand for loops in python better!

Friday, 13 June 2014

First Week!

      Started putting together a series of structures for ILE using python ... I think I will never understand how for loops and dictionaries work! Oh well its good fun and sort of shows how useful computing can be since now its easy to produce any number of ILE rotamers.

Towards the end of this week I managed to measure the phi/psi angles for the 'top500' PDB files. Hopefully these will be able to give me an indication of the backbone angles for that are populated so that I can make chi 1 and ch 2 rotamers for each one.