search

3dem / model-angelo

Automatic atomic model building program for cryo-EM maps
MIT License
124 stars 19 forks source link

question about using the model-angelo with no sequence #110

Closed PeterVanF closed 1 month ago

PeterVanF commented 2 months ago

Hi, I have recently been using Model-Angelo to identify an unknown density determined by cryo-EM with a local resolution of about 4 Å. I'm uncertain whether this region belongs to RNA or protein. Despite the results being incomplete and fragmented, I used Model-Angelo (no-sequence) to build its atomic model. Some regions were identified by Model-Angelo as nucleoside bases, while others were identified as amino acids.

I would like to know if these results are reliable. Are there confidence or probability values available that can help me analyze this structure more accurately?

jamaliki commented 1 month ago

Hi,

If you run the program with the flag "--keep-intermediate-results" then it should save extra files into directories named "gnn_outputround*". If you look at the last folder, "gnn_output_round_3", there should be a file called output_aa_probabilities.aap which gives you the probability of every residue (in order) for every class. You can then see what the relative probabilities between protein and nucleotide are.

Best, Kiarash.

PeterVanF commented 1 month ago

I got it. Thank you!