I am trying to hone my de novo design skills by reproducing de novo
designs of Top7 as described by Kuhlman, et al. (2003). So far, I can
generate sequence-independent starting models that I am confident are
good, but I am clearly not doing well on the flexible backbone design
step. After flexible backbone design, my best per residue LJ
attractive energies are around -3.1 per residue, whereas the same
value reported in Table S1 from Kuhlman, et al. (2003) is about -4.1
per residue (-385 / 93 resis). I also frequently get structures that
'blow up'; they are no longer compact globules, but very extended
My resfile is set up to allow only polar amino acids on the
solvent-exposed side of the beta sheet, and everything but cysteine
everywhere else, as described for Top7.
I have tried flexible backbone design without a weights file (default
weights), or giving Rosetta the weights listed in Table S6 from
Kuhlman, et al. (2003) both with and without the unique reference
energies for each amino acid in a -weightfile. I find that when I use
the reference energies listed in that table I get sequences extremely
heavy in E, D, and K, which means that the reference energy term is
dominating everything else.
Here is the command that I am running:
rosetta.gcc XX ABCD A
My results clearly suggest that I am not doing something right, but I
can't figure out what it might be. Any help would be greatly
appreciated. Thanks in advance!