First, thank you for all the time and effort put into making OmegaFold such an aswesome tool.
We are implementing OmegaFold as part of an AI-based pipeline for NMR resonance assignment and we would like to force OmegaFold to produce low quality models to be used as decoys (to test the robustness of the pipeline in those cases in which OmegaFold encounters a sequence poorly described by its training dataset). So, I would like to kindly ask:
Is there any way to make OmegaFold yield low-quality models mimicking the behavior of facing sequences poorly described by the model?
Alternatively, is there any way to make OmegaFold compute the pLDDT on a user-provided pdb file (or any other scoring function that could be used to rank conformations)? This way, we could create random conformations and rank them with OmegaFold to get a sample of increasingly worse decoys.
Thank you very much for any insight you could offer on this.
Hi!
First, thank you for all the time and effort put into making OmegaFold such an aswesome tool.
We are implementing OmegaFold as part of an AI-based pipeline for NMR resonance assignment and we would like to force OmegaFold to produce low quality models to be used as decoys (to test the robustness of the pipeline in those cases in which OmegaFold encounters a sequence poorly described by its training dataset). So, I would like to kindly ask:
Is there any way to make OmegaFold yield low-quality models mimicking the behavior of facing sequences poorly described by the model?
Alternatively, is there any way to make OmegaFold compute the pLDDT on a user-provided pdb file (or any other scoring function that could be used to rank conformations)? This way, we could create random conformations and rank them with OmegaFold to get a sample of increasingly worse decoys.
Thank you very much for any insight you could offer on this.
Kind regards.