search

aaayushg / generative_IDPs

Autoencoder based model to mine new conformations of IDPs
2 stars 4 forks source link

name 'decoder' is not defined in autoencoder_train_generate.py line 245 #1

Closed wowfee closed 1 year ago

wowfee commented 2 years ago

hello, when i run python autoencoder_train_generate.py input after finished autoencoder model created! step it returned error : name 'decoder' is not defined in autoencoder_train_generate.py line 245 i check the autoencoder_train_generate.py ,the decoder is not defined, could you please give me any suggestions ?

aaayushg commented 2 years ago

Thanks for trying the code and reporting the error. I just corrected the code: decoder needs to be defined "decoder = Sequential()" on line 245.

wowfee commented 2 years ago

Thank you for your help.it work now! i also wonder to know wether this method could be used to predict the conformation of a protein with hinge region and well fold part ?

aaayushg commented 2 years ago

Great. I'm glad that it works for you. Yes, this model will work on variety of protein loops. It'll be interesting to evaluate this architecture performance on different systems.

wowfee commented 2 years ago

hello, i have another question,after run autoencoder_train_generate.py. i get three pdb files: test.pdb decoded.pdb and cov.pdb, i know test.pdb is the remainder in the long trajectory (ig. 80% of a trajectory for 20% training case) i wonder to know the difference of decoded.pdb and cov.pdb ? and which one is the prediction structure?? thank you very much!

aaayushg commented 2 years ago

Hi, Thanks again for trying out our code. decoded.pdb is reconstruction of test set. cov.pdb is new generated set (sampled from gaussian) or newly predicted structures.

Let me know if you have any questions.

Aayush

On Wed, Jun 8, 2022 at 8:11 AM feijunwen @.***> wrote:

hello, i have another question,after run autoencoder_train_generate.py. i get three pdb files: test.pdb decoded.pdb and cov.pdb, i know test.pdb is the remainder in the long trajectory (ig. 80% of a trajectory for 20% training case) i wonder to know the difference of decoded.pdb and cov.pdb ? and which one is the prediction structure?? thank you very much!

— Reply to this email directly, view it on GitHub https://github.com/aaayushg/generative_IDPs/issues/1#issuecomment-1149390928, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALKXTF6U3NKDOVV6GKP3GR3VOAB6ZANCNFSM5LNV5KXQ . You are receiving this because you commented.Message ID: @.***>

--

Aayush Gupta, PhD Research Scientist, AI/Comp-Chem Exscientia/University of Illinois at Chicago Website: aaayushg.github.io http://aaayushg.github.io

wowfee commented 2 years ago

thanks for your help! Through visualization, I found that the cov structure has serious deformation problems, and it is difficult to use it later. Is there a good way to deal with this irrationality of the cov structure?

aaayushg commented 2 years ago

Yes, you can minimize all the structures using FF. We tried it and it works. We also updated our work which will be published soon based on this.

On Wed, Jun 8, 2022 at 8:26 AM feijunwen @.***> wrote:

thanks for your help! Through visualization, I found that the cov structure has serious deformation problems, and it is difficult to use it later. Is there a good way to deal with this irrationality of the cov structure?

— Reply to this email directly, view it on GitHub https://github.com/aaayushg/generative_IDPs/issues/1#issuecomment-1149398765, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALKXTF7OP5MO7M6SXI3BO4TVOADW3ANCNFSM5LNV5KXQ . You are receiving this because you commented.Message ID: @.***>

--

Aayush Gupta, PhD Research Scientist, AI/Comp-Chem Exscientia/University of Illinois at Chicago Website: aaayushg.github.io http://aaayushg.github.io

wowfee commented 2 years ago

thanks, I'll try to use minimization and look forward to the publication of your work. thanks again for your help