Cbioportal: 3d structure of tp53 not showing up

[inodb]

Might be a problem with multiple genes

[jjgao]

@onursumer any update?

[onursumer]

@jjgao @inodb Looks like we rank 5XZC as the highest PDB structure to load. However, 3Dmol cannot download any structure for this PDB: http://3dmol.csb.pitt.edu/viewer.html?pdb=5XZC.

[onursumer]

NGL has no problem loading it: https://www.rcsb.org/3d-view/5XZC/1, maybe we should switch to NGL :)

[onursumer]

Opened a ticket at 3dmol.js issue tracker: https://github.com/3dmol/3Dmol.js/issues/415

[inodb]

@onursumer

• might be an option. Im not too familiar with either. If the general consensus is that NGL works better these days it could be worth switching. Does seem like it could potentially cause other issues
• how does the current ranking of PDB structures work? What is it based on?

[onursumer]

@inodb ranking is based on the identity value and the alignment length returned by g2s service. Here you can find the full sort metric: https://github.com/cBioPortal/cbioportal-frontend/blob/9c5b2e874d5d08a9a46a10521c2b5d8918dea6d8/src/shared/components/mutationMapper/MutationMapperStore.ts#L371

[inodb]

@onursumer Thanks! If there's no quick response to the 3d mol issue tracker bug: maybe a temporary fix could be to have a list of ignored structures?

[onursumer]

@inodb that may work, but detecting empty structures might be tricky/hacky. The only indication is this webGL warning message in the console: [.WebGL-0x3c66aae86000]RENDER WARNING: Render count or primcount is 0.

[inodb]

@onursumer good point - i thought maybe we only hardcode the TP53 structure 5XZC. Then add more if any other broken ones are reported. Until we find a proper fix. But yeah let's see if it's an easy fix on their end

[jjgao]

NGL is getting popular. PDB has also adopted it. I am not sure how much work it would be to switch?