how to build a dataset of random peptides from naturally occurring proteins

Question

[dr.f.liu]

Hi, guys,

I am working on MHC binding prediction, and a dataset of random natural peptides (length 8-11aa, 20,000-1,000,000 in number) is needed to build a rank-score table.

I have checked the papers on netMHC, netCTL and IEDB analysis resources, but they didn't describe their methods to generate the random peptides in detail.

One of possible solutions is to generate random peptides from protein sequences in SwissProt, but still not start yet.

Do you have any idea of this question?

Thank you in advance.

[Nitin Narwade]

I think one possible solution for this could be; download peptides sequences from IEDB database then you can generate random peptides sequences using the same as a template with different permutations and combinations. At the end remove the duplicates from randomly generated peptides.