RNA 3D coordinates

[taywpd]

Dear Dr. Oliver,

Thank you for the very nice program. I was wondering:

Assuming we have a predetermined small secondary RNA structure we want to designate as the binding pocket, is there a way to convert 3D coordinates of that RNA secondary structures into the NetworkX graph input required by the RNAmigos model?

Cheers,
Dillon

[cgoliver]

Hello Dillon,

Thank you for your insterest in RNAmigos.

This is always the trickiest part of the whole thing.

First, I assume that by secondary structure you mean tertiary structure since for this to work you need to have 3D coordinates of the RNA and not just the secondary structure (canonical base pairs).

It all depends on whether or not the RNA you are using is in the RCSB PDB Databank or not.

Case 1.: In the PDB Databank

• In the original paper we used annotations generated by FR3D this is a dataset that automatically annotates all RNAs available on the PDB. If yours is available on the PDB then you can use this option.
• We built a library to help with these graphs called rnaglib where you can easily load the graph for any RNA in the PDB.

Case 2: Not in the PDB Databank

• Again through rnaglib, if you have x3dna-dssr installed you can build the graph with this function
• Another option is there is a new library called barnaba which is open source and runs in python so you can feed it a PDB and it generates the necessary annotations, from that it's straightforward to build a graph. In the coming days we will integrate this into rnaglib so you can have it all done in one place.

tldr; it's easy if your RNA is available in the PDB, if not, there is some work to do but it's not impossible. In both cases, I'm happy to help.

Best,
Carlos