incomparable ddG values

Hello,

Could you help me, please?
I need to compute 30 single mutations. The job with all 30 single mutations in a single mutfile fell down with an incomprehensible comment. Therefore, I have tested single mutfiles containing 2, 4, 8, and 12 single mutations. It has been worked, but energy values have been incomparable.
When I had used the file with 2 single mutations in the single mutfile I got these values:
WT:   864.218;  MUT_8GLN:   864.873; MUT_10LYS:   864.274.  
For 4 single mutations
WT:   749.502;  MUT_8GLN:   749.788; MUT_10LYS:   748.841; MUT_16LYS:   747.401; MUT_25LYS:   749.223.
For 8 single mutation
WT:   643.821; MUT_8GLN:   643.757; MUT_10LYS:   642.312; etc.
The complete output files for 8 and 12 single mutations you can find as attachments.

In my opinion, the energies should be the same for the same WT and the mutations, of course, except for decimal places because of the procedure of the minimization.  

I have computed the change of stability, ddG = dGmutant ‐ dGwildtype, with Rosetta 3.9. The protocol 'cartesian_ddg application' was used, see below. The crystal structure was adjusted by the Rosetta module 'clean_pdb.py' and by the recommended pre-minimization. The protein includes two disulfide bridges.
Input.mutfile was prepared in this way:
total 2
1
R 60 Q
1
E 62 K
etc.

Options:
-in:file:s dep2_0020.pdb
-ddg::mut_file dep2.mutfile
-database /software/rosetta/3.9/main/database
-fa_max_dis 9.0
-ddg:iterations 3
-ddg::dump_pdbs false
-ddg::cartesian
-bbnbr 1
-score::weights /software/rosetta/3.9/main/database/scoring/weights/score12_cart.wts
-ddg::out

Thank you in advance for your answer.
Best regards,
Ela