sokrypton
Can I use PDB70 instead of PDB100?
Current Behavior
Hi, I have been running ColabFold v1.5.2-patch to predict a protein complex. I have noticed that the predicted structure I was getting with the previous PDB70 database is much better than the one I am getting with the current PDB100 database. Is there a way to run this prediction with PDB70 now? Or could this change be due to some parameter change? I have been using the default settings. I have already tried both complete and greedy pairing.
Context
I have some new analyses to do and I would like to do it in the same way as I did before ( before the database got updated to pdb100) which gave me a better prediction.
Your Environment
Include as many relevant details about the environment you experienced the bug in.
- Git commit used
- If you run it on a local system. Please add the server specifications
- Operating system and version:
I agree with you,. I have the same problem to finish some analyses. But I think other parameters have been changed. Have a look into the file attached to see differences on the same job. Results.pdf
@cdelam The advanced notebook is actually using a hacked version (with residue index offset) of the monomer model to make the predictions for the multimer. Within the new notebook, try set: model_type: alphafold2_ptm
@sokrypton Thank you very much. This is very helpful. Do you know if there is any documentation about these different model types?