search

barthelemymp / TULIP-TCR

GNU General Public License v3.0
10 stars 3 forks source link

Directly comparing wtih different peptides to a given TCR #14

Open CNDOTA opened 3 months ago

CNDOTA commented 3 months ago

Dear authors, congratulations on your very nice work!

I have one problem.

Could I use the TULIP model to compare the binding strength of different peptides (such as random peptides) given a fixed TCR?

Thank you very much!

barthelemymp commented 1 month ago

Hi,

Sorry for missing your issue. Changing the peptide for a fixed TCR is harder to compare. The current default score, (peptide log likelihood) will not work. You could have a look to the MI on both chain maybe ? Problem is that you have two binding at plays here tcr peptide and peptide mhc. Another alternative would be to use a repertoire, for each peptide you rank you evaluate your tcr + all the rep. you use the rank of your TCR inside the repertoire as a score.

This would need some little test to check if it works.