Repository navigation
Home
We use weighslide to understand and optimise data related to alpha helices in proteins. The helix repeats every 3.6 residues. In a sequence of residues "ABCDEFG", after being wound into a helix residue A will be closest (in space) to D and E. Each residue will have certain properties (hydrophobicity, size, conservation, etc). A primary protein sequence can therefore be represented as a 1D array of property values.
example question: "Which face of the helix is the most hydrophobic"?
method: create a 1D array of hydrophobicity values, run weighslide with the desired window that averages over a helix face ("9xxx9xxx9", or "xxxx9xxx9", etc), and find the window with the highest value
We also need to process 1D data to analyse or predict interfaces of interacting helices. Commonly, the interface of left-handed helix pairs is thought to follow a heptad motif (a, de and g of "abcdefgabcdefg"), whereas the interface of right-handed helix pairs is thought to follow a tetrad motif (ab of "abcdabcdabcd").
To experiment with various windows, it makes sense to have a flexible program.
Weighslide has only been tested on small 1D datasets. Currently an output excel file with all windows and a figure of the data is automatically created. The performance will be slow. If you want to try it, grab the code, alter it so the outputs are optional, and create a pull request.
You should check first if the python scipy signal module has a window that suits your needs.
http://docs.scipy.org/doc/scipy-0.14.0/reference/signal.html
Currently scipy has implemented the following windows (which for some unknown reason all seem to start with the letter "b"):
barthann, bartlett, blackman, blackmanharris, bohman, boxcar, chebwin, cosine, flattop, gaussian, general_gaussian, hamming, hann, kaiser, nuttall, parzen, slepian, triangle