RuntimeError: shape '[16, 20]' is invalid for input of size 240

[ricosuavee]

I'm trying to determine a structure of an unknown contaminant protein by cryoEM single particle analysis. I wanted to generate a main-chain model of my density map from cryoSPARC. However, I get the following error (on both flipped unflipped density maps in .mrc format) during GNN model refinement Round 2/3:

RuntimeError: shape '[16, 20]' is invalid for input of size 240

I am running modelAngelo with 8 CPUs, 1 GPU, and 60 GB of RAM on a cluster (submitted with Sbatch), using the following script (SBATCH submission lines excluded):

module load Anaconda3/ module load CUDA/11.1.1-GCC-10.2.0 source activate model_angelo model_angelo build_no_seq -v /path/to/mrc.mrc
-o output/path source deactivate

I have previously had success (no error) running this same script on a lower resolution density map (the output was junk because map was low resolution, but no error in the log).

Full error traceback attached from model_angelo.log model_angelo_error.txt

[jamaliki]

Hi!

Sorry it seems that ModelAngelo has found less than 20 residues, so it results in this bug. Is that what you are expecting? It might be that the resolution is still too low for ModelAngelo to find enough residues.