search

patrickbryant1 / SpeedPPI

Rapid protein-protein interaction network creation from multiple sequence alignments with Deep Learning
Other
65 stars 14 forks source link

protein ordering changes the pdockQ score #26

Open akdess opened 1 month ago

akdess commented 1 month ago

Hi,

Thanks a lot for developing this great tool.

I am trying to run interaction analysis on two proteins.

bash predict_single.sh A.fasta B.fasta gives pdockQ score of 0.5 bash predict_single.sh B.fasta A.fasta gives pdockQ score of 0 How should I interpret this score?

Additionally running bash predict_single.sh A.fasta B.fasta also sometimes give inconsistent results, sometimes higher than 0.5 and sometimes around 0. It might be related to the Alphafold algorithm random features.

Thanks a lot!

patrickbryant1 commented 1 month ago

Hi,

This is interesting. Yes, the MSAs are samples which can result in different results, but it should not differ this much on average. Can you please run it 5 times in each order and confirm that the scores remain the same?