A question about the result of protein-DNA complex analysis

After I run the Haddock of protein-DNA complex analysis. I found difference between before and after the results. Now
I wonder how many differences are acceptable in the Z-Score results.

The server reports the clusters sorted according to the HADDOCK score. The score is calculated as the average over the top four members of each cluster. If scores if different clusters are overlapping you will have to find other information to select one or another.

And what to you mean by “I found differences before and after the results”???

Thanks for your answer.
Such as the input of exactly the same data, the first time the average score of my best clusters is -1.9, and the second is -2.3.I wonder whether the deviation of 0.4 can be accept or not?

I would rather compare the HADDOCK scores and their standard deviations.

These calculations are chaotic by nature. Your simulations are run on the EGI grid, meaning that the computations will run on different hardware. This can influence the results - but ideally the clusters should still be similar, even if the exact energetics numbers might be a bit different.