search

KarchinLab / bigmhc

BigMHC predicts MHC-I (neo)epitope presentation and immunogenicity
Other
45 stars 9 forks source link

about output results #3

Closed Xiaoguang-Shi closed 11 months ago

Xiaoguang-Shi commented 1 year ago

Hello, thank you for providing this tool. I have some confusion regarding the interpretation of the output results. How can I understand the affinity and immunogenicity from the BigMHC_EL and BigMHC_IM results? Is there a reasonable cutoff value?

benjaminalbert commented 1 year ago

Hi @Xiaoguang-Shi ,

The scores that BigMHC produces are in the range (0,1) but it is primarily intended to sort a list of putative neoepitopes. For example, if one has the capacity to validate N neoepitopes per patient in vitro, then one would rank all neoepitopes (potentially stratifying by allele first) and take the top N to validate. From those N that are validated, a subset that generated an in vitro immune response may be usable in an immunotherapy. From this pipeline, the "cutoff" is dictated solely by the capacity for experimental validation.

If for some reason you still want to approximate an output threshold, you could estimate that number based on ROC and/or PR curves for a metric of interest (e.g. choose the threshold that preserves approximately X precision).

Xiaoguang-Shi commented 1 year ago

Thank you for your reply.I will try according to your suggestion.

benjaminalbert commented 1 year ago

Hi @Xiaoguang-Shi, I wanted to follow up and see if this is still an issue?

zhangkn3 commented 8 months ago

Hi Benjamin, Thanks for the development of the software. I found the BigMHC IM range from 0 to 1. Will the neoantigen candidate be more immunogenetic when the value of BigMHC IM is higher or lower?

benjaminalbert commented 8 months ago

Hi @zhangkn3, higher output indicates greater immunogenicity.

zhangkn3 commented 8 months ago

Thanks for your quick response!