Zhanglab forum

Hi, everyone,
I recently installed DMFoldv1.1 to predict multimer proteins. I rerun the example1 file. And I got "DMFAF2 jobs are not complete yet" end in Run_DMFold.py .
I found there were 40 joint MSAs in the AF2Models which shoud be used to predict. For each MSA, there were 5 AF2 models, and for each AF2 models in each MSA, AF2 would predict 5 times. SO the final models supposed to be 40*5*5=1000
But I only successfully got 625 .pdb file in DMFoldModels. So it meant only 25 joint MSAs were used. It's same in your example file.
So, are these 25 MSAs the first 25 ranked by sepecific method?
In fact, when i found the end info in nohup file but no DMFoldModels directory. I tried to run Run_DMFold.py again. And Then I got DMFoldModels directory. It's bug?

Hope someone can answer my questions! Thanks!

And finally thank you guys for this powerful tool

BW
Xilong

Dear Xilong,

To speed up the modeling process, we fine-tune the maximal modeling MSA pair as 25 (scoring MSAs by M-score, see the paper for details), thus you get 25*5*5=625 models. The results you get are complete. While, if you want to generate more models, you can enlarge the cutoff from 25 to more (such as 40) in the config.py in the package. Let me know if you have any questions.

Best Regards
Wei Zheng