ProtSSN icon indicating copy to clipboard operation
ProtSSN copied to clipboard

multiple mutation

Open polytide1988 opened this issue 3 months ago • 2 comments

  1. Just try the compute_fitness.py and find the output in .../result/ is the score of single mutations like the attached one. But now, we hope to see the score of something like that : Q1V;D3V;A6K..... i.e. the score of several mutations combined together. Can ProtSSN achieve that ?

1A7V.csv

  1. In https://github.com/ai4protein/ProtSSN, section "Test on Your Own Dataset", the zeroshot_predict.py is mentioned but at the top of this github, you mentioned that (https://github.com/ai4protein/ProtSSN/commit/ace6a0bd3aaca9ac2f856839a777f9527b5f4db0) the zeroshot_predict.py was replaced by compute_fiteness.py Suggest you make the corresponding change in "Test on Your Own Dataset" :)

polytide1988 avatar Aug 29 '25 00:08 polytide1988

  1. Sure. We support multi-mutation prediction.
  2. To fit the format of ProteinGym, we updated the predict script name to compute_fitness, but they are the same.

tyang816 avatar Aug 29 '25 05:08 tyang816

Just put the multiple mutant into the data/mutant_example/DATASET as .csv and .tsv file and re-run the compute_ fitness.py and generate the attached result/DATASET/FGF2.csv . After row 2776, some multiple mutants are there as attached. Now, our questions is : How is your calculation logic when multiple mutations are applied ? Since, multiple mutations are not the simple story of adding scores of single mutation together. Since, 1+1 might not equal to 2 in protein world.

FGF2.csv

polytide1988 avatar Aug 29 '25 23:08 polytide1988