In article <ghermans-0404971321560001 at pstn11.luc.ac.be>, ghermans at luc.ac.be (Guy Hermans) writes:
> In article <5ht3ja$odd at rc1.vub.ac.be>, gbottu at ben.vub.ac.be (Guy Bottu) wrote:
>>> Does someone know of a software to predict T cell epitopes or
>> maybe a public Server where a protein can be submitted to such
>> a program ?
>>>> Dr. Guy Bottu
>> A number of methods has been proposed and evaluated. One Mac program,
> called T-sites, incorporates a number of such methods.
>> It is freeware, and new versions could be obtained from the company where
> the authors worked _at the time_. I tried to contact them for a new
> version, but to no avail.
>> I'll include the program in a personal e-mail to you, so I won't have to
> overload newsservers all over the planet. If someone out here wishes to
> get a copy, please let me know personally.
>> Greetings to all
You can be just as lucky if you write the protein sequence on the piece of
paper, close eyes and point finger somewhere on the written sequence. There
are almost two hundred known human class I MHC molecules, even more class II
molecules. The likelihood is that every, say, 15-mer peptide from within a
protein will bind to some of the MHC variants and therefore be a candidate
for a T-cell epitope.
Prediction of peptide binding to a particular MHC molecule can be done and
has been done reasonably well for HLA-A*0201, -A2401/02, -B2705, -B3501 and
-DRB1*0401. Claims for solutions including some other alleles exist, but
whether they were properly evaluated is unclear.