Problem in redesign binding site

[Anonymous]

Hi everyone,

Now, I am studying about protein design in carbohydrate modifying enzyme. I am very new for Rosetta and trying to use Rosetta Design to change the product specification. I have constructed intermediate molecules in catalytic step and docked into the active site of the enzyme by another program. Then I performed protein design by using fixbb and enzyme_design (without constraint) applications to redesign residues in the binding site around intermediate molecules. But they seem like they could not see intermediate molecules. Both fixbb and enzyme_design applications changed the residues in the binding site to interact with themselves instead. Here is an input files, does it has any option or command that I miss to make an interaction or any program would fit to my purpose?

Any advice would be much appreciated

 

[Anonymous]

I’m not seeing anything majorly wrong with your input options files. (Though I would highly recommend removing -constant_seed, unless you’re also explicitly changing the seed value with -jran. Also, the -dun10 flag should be superflous with recent versions of Rosetta – BTW, which version of Rosetta are you using?)

I’m not quite understanding the symptoms you’re describing. Does the ligand show up in the output structures? That is, are the atoms for the ligand present? (If not, do you have the ligand in the input structure?) Are the ligands of the appropriate structure and in the appropriate locations? Or is it that you don’t think the design has sufficient protein-ligand interactions designed in?

If the last, one thing to keep in mind is that even when designing ligand binding, Rosetta is optimizing the energy of the entire structure, so if Rosetta thinks it can get more energy by forming new interactions within the protein versus between the protein and the ligand, then it will do that. There’s several approaches you can take to limit/avoid this. One of the easiest might be to identify which residue identities are subject to this self-interaction, and then change your resfile to avoid those particular residue identities. Another approach, if you’re using the enzdes protocol, is to use the protein-ligand interface upweighter. This multiplies the effect of protein-ligand interactions specifically at designed positions, to encourage Rosetta to put in protein-ligand interactions. You can enable this by adding the option -enzdes:lig_packer_weight which takes as a parameter the factor by which to upweight the protein-ligand interactions. (e.g. “-enzdes:lig_packer_weight 2.0”) Typical values are 2.0 to 5.0. If you have to go more than that to see an effect, then there’s probably something else going on, and the interaction upweighter is simply masking the issue.

[Anonymous]

I’m not seeing anything majorly wrong with your input options files. (Though I would highly recommend removing -constant_seed, unless you’re also explicitly changing the seed value with -jran. Also, the -dun10 flag should be superflous with recent versions of Rosetta – BTW, which version of Rosetta are you using?)

I’m not quite understanding the symptoms you’re describing. Does the ligand show up in the output structures? That is, are the atoms for the ligand present? (If not, do you have the ligand in the input structure?) Are the ligands of the appropriate structure and in the appropriate locations? Or is it that you don’t think the design has sufficient protein-ligand interactions designed in?

If the last, one thing to keep in mind is that even when designing ligand binding, Rosetta is optimizing the energy of the entire structure, so if Rosetta thinks it can get more energy by forming new interactions within the protein versus between the protein and the ligand, then it will do that. There’s several approaches you can take to limit/avoid this. One of the easiest might be to identify which residue identities are subject to this self-interaction, and then change your resfile to avoid those particular residue identities. Another approach, if you’re using the enzdes protocol, is to use the protein-ligand interface upweighter. This multiplies the effect of protein-ligand interactions specifically at designed positions, to encourage Rosetta to put in protein-ligand interactions. You can enable this by adding the option -enzdes:lig_packer_weight which takes as a parameter the factor by which to upweight the protein-ligand interactions. (e.g. “-enzdes:lig_packer_weight 2.0”) Typical values are 2.0 to 5.0. If you have to go more than that to see an effect, then there’s probably something else going on, and the interaction upweighter is simply masking the issue.

[Anonymous]

I’m not seeing anything majorly wrong with your input options files. (Though I would highly recommend removing -constant_seed, unless you’re also explicitly changing the seed value with -jran. Also, the -dun10 flag should be superflous with recent versions of Rosetta – BTW, which version of Rosetta are you using?)

I’m not quite understanding the symptoms you’re describing. Does the ligand show up in the output structures? That is, are the atoms for the ligand present? (If not, do you have the ligand in the input structure?) Are the ligands of the appropriate structure and in the appropriate locations? Or is it that you don’t think the design has sufficient protein-ligand interactions designed in?

If the last, one thing to keep in mind is that even when designing ligand binding, Rosetta is optimizing the energy of the entire structure, so if Rosetta thinks it can get more energy by forming new interactions within the protein versus between the protein and the ligand, then it will do that. There’s several approaches you can take to limit/avoid this. One of the easiest might be to identify which residue identities are subject to this self-interaction, and then change your resfile to avoid those particular residue identities. Another approach, if you’re using the enzdes protocol, is to use the protein-ligand interface upweighter. This multiplies the effect of protein-ligand interactions specifically at designed positions, to encourage Rosetta to put in protein-ligand interactions. You can enable this by adding the option -enzdes:lig_packer_weight which takes as a parameter the factor by which to upweight the protein-ligand interactions. (e.g. “-enzdes:lig_packer_weight 2.0”) Typical values are 2.0 to 5.0. If you have to go more than that to see an effect, then there’s probably something else going on, and the interaction upweighter is simply masking the issue.

[Author]

Posts