issues about repeatability

[Kange2014]

Hi,
it's nice to see T-SCAPE here for immunoginicity prediction. however, during my test, I found two questions that hope to get clear from you:

1. if the same peptide with p_im type was predicted twice using the same command, it would produce differenct scores . Can u help explain why?
2. I also tried to score the FDA antibody ADA benchmark dataset, using p_im type inference based on the introduction of 'ADA level prediction of protein drug sequences' part. However, I found most of antibodies including chimeric and mouse antibodies are scored below 0.2 (as the figure shows). This seems to be different from what you claimed in the paper.

Look forward to your kind feedback.

thanks.

[vinbl]

Fixed the issue, see here