Hi all,
I have used PSI-Pred to generate secondary structure predictions of each scanned alanine mutation. How do I compile all this data to look for significant deviations from the wild-type protein?
My approach was to use a sliding window (say 5 amino acids wide) and take a weighted average with respect to the centre amino acid. From this, I wasn't sure how to proceed. Maybe I could take all the averages from matching window positions and use the average to run a t-test against the wild-type.
Can you impart some wisdom?
•
written
by
leonardo •
I would do what you did. Align each prediction to the wild type and see where the secondary structures are more conserved throughout the scans. I would also look at regular conservation and see where you have the most conserved residues and pay more attention to those regions.
Do you have structures or can you build a model for the protein? That would be much more reliable (then again, depends on the model).
I can look at conserved residues to see if these regions are more affected than non-conserved domains. There are no crystal structures yet. Is there a more objective or statistical way to compare differences?
Log in to answer this question.
