# RNA modeling with step wise Monte Carlo error

4 posts / 0 new
RNA modeling with step wise Monte Carlo error
#1

Hi Rosetta team,

I'm trying to use step wise Monte Carlo to model RNA structure.

Here is the command I used:

 stepwise.static.linuxgccrelease -fasta 2gdix.fasta -s H0_minimize.pdb H1_minimize.pdb H2_minimize.pdb H3_minimize.pdb H4_minimize.pdb H5_minimize.pdb -input_res 1-5 76-80 6-9 39-42 12-14 27-29 16-18 24-26 48-50 71-73 54-57 64-67 -nstruct 10000 -cycles 10000 -out:file:silent swm_2gdi_nores.out -motif_mode -minimize_rna -allow_complex_loop_graph true -in:file:native 2gdix_RNA.pdb -cst_file fade_0_50_cd10.cst -score:weights stepwise/rna/rna_res_level_energy4.wts -restore_talaris_behavior


I get a strange error and the job just seems to fall into an infinite look without producing an output file:

[FILE]: src/core/pose/full_model_info/FullModelInfo.cc
[LINE]: 372
[START_MESSAGE]
[ ERROR ] UtilityExitException
ERROR: Assertion idx == 0 failed.


I wonder if it's something with the script? I attached part of my crash file as well. Thank you.

Sincerely,

Dan

AttachmentSize
8.36 KB
Post Situation:
Mon, 2020-12-07 18:08
szcshr123

I'm so happy you're using our code! It'll be tricky to figure out the reason for your error without seeing your input files and fasta.  The most likely reason for an issue is an overlap between the files, or something else going wrong. If you're setting up the TPP riboswitch specifically (I recognize 2GDI?) then I can set up my own input files and show you how I would approach the command line, if you'd like that?

Mon, 2020-12-07 18:32
everyday847

Thank you for the timely reply. Yes it is TPP riboswitch 2GDI. I attached my fasta and constraint file.

File attachments:
Tue, 2020-12-08 04:37
szcshr123

I maybe wasn't specific enough: I think you need to share your input helices, in particular, to make sure they correspond to the FASTA. In particular, I'm concerned that your fasta is just a bare sequence line. That's not our standard format; you might want something like:

>2gdi A:X:10-89
ggacucggggugcccuucugcgugaaggcugagaaauacccguaucaccugaucuggauaaugccagcguagggaaguuc

if you wanted to match exactly the format of chain A (segment ID X) of the PDB 2gdi. And your input helices should match that nomenclature as well. If you want to cheat a little (to save yourself a little time initially -- not as a fully blind test!) you can save them straight out of the native PDB; otherwise, you can use rna_helix to generate ideal ones.

Anyway, that's a really big modeling problem! Usually we use something like FARFAR2 for whole RNAs, and SWM just for improving individual junctions -- unless we can reduce the modeling problem to something sort of "motif-sized." I'm very interested to hear how it goes!

Wed, 2020-12-09 21:46
everyday847