Extracting annotations from structures

[Ninjani]

By: Barbara Terlouw, Mehmet Akdel, Janani Durairaj

These are the annotation types/sources that we are planning to work on. We'll be writing scripts such that the annotations can be generated for any PDB file so that it works for new ones as well. These annotations are mostly per structure, per residue, except for the first two. Suggestions welcome!

• functional domains
• phenotypes
• point mutations
• hydrophobicity
• surface accessibility
• residue fluctuations derived from elastic network models (ENMs)
• perturbation response from ENMs
• mechanical stiffness from ENMs
• hinge sites from ENMs
• RMSD of residues across structures of the same protein with different ligands bound (this approach can easily be applied to compare COVID proteins to SARS or other related proteins instead)

[gtauriello]

@all-contributors please add @BarbaraTerlouw for ideas

As I expect this effort to include contributions from several people, please feel free to acknowledge them using the all-contributors bot at described in the contribution guidelines

[allcontributors]

@gtauriello

I've put up a pull request to add @BarbaraTerlouw! 🎉

[Ninjani]

Added a notebook https://github.com/BarbaraTerlouw/covid19/blob/master/Ensemble.ipynb to analyze the variability in RMSD between the different versions of the SARS-Cov2 3CL-PRO protein (with different ligands bound), and a SARS-Cov 3CL-PRO protein.

[akdel]

Update on progress:

• functional domains
• phenotypes
• point mutations
• hydrophobicity
• surface accessibility
• residue fluctuations derived from elastic network models (ENMs)
• perturbation response from ENMs
• mechanical stiffness from ENMs
• hinge sites from ENMs
• RMSD of residues across structures of the same protein with different ligands bound

The annotations are not yet converted to the SWISS-model format.

Example annotations on 3CL-pro (main protease):

• anm fluctuations
• perturbation effectiveness
• perturbation sensitivity
• mechanical stiffness
• rmsd per residue compared all SARS-cov2 main protease structures
• pca fluctuations compared all SARS-cov2 main protease structures

To be able to see each of these annotations:

1. Click on the protein ID (P0DTD1) after opening one of the above links
2. Choose the 3C-like proteinase
3. Click on annotations from "Covid-19 Annotations" box

[gtauriello]

Nice work. Thanks. On our end we are discussing display issues since by default per-residue annotations get a ball-and-stick display which doesn't really work well for your type of data. As a temp. workaround for the NGL 3D viewer: if you select "Cartoon" view after selecting the annotations you can get rid of the sticks.

[mlgill]

@all-contributors please add @mlgill for solvent accessibility and interface annotations

[allcontributors]

@mlgill

I've put up a pull request to add @mlgill! 🎉

[akdel]

We were thinking of representing possible conformational movements for each protein by using:

1. Information from multiple existing structures
2. Predictions from elastic network models

Would this be a good idea?

They can be represented either using arrows on the models or by animating the model:
Spike protein

3CL-protease

[mlgill]

@akdel Do we know the direction of movement from your calculations? I've seen it done by varying the width of the backbone for NMR spin relaxation measurements, etc.

[mlgill]

Greetings all, new to the hackathon and group. I introduced myself in the channel if you're looking for more info.

Per discussion here I'm going to start working on the solvent accessibility calculations.

WIP pull request is #34 to track my work

[akdel]

@akdel Do we know the direction of movement from your calculations? I've seen it done by varying the width of the backbone for NMR spin relaxation measurements, etc.

The direction is stored per residue for a given time in normal mode data format. Some more context here. Did I understand your question correctly?

[mlgill]

@akdel Do we know the direction of movement from your calculations? I've seen it done by varying the width of the backbone for NMR spin relaxation measurements, etc.

The direction is stored per residue for a given time in normal mode data format. Some more context here. Did I understand your question correctly?

Yes, that's my question. Thanks -- I was trying to understand how specific the directionality was or if it was more indicative of the magnitude of motions. Sounds like the former.

[mlgill]

Have pushed a draft of the code to extract surface accessibility values and write a CSV for a given PDB. What else is needed? Feedback welcome.

#34

Thanks.

[Ninjani]

Very cool! We can pull your code and add a small function to generate a SWISS-MODEL annotation URL that visualizes the accessibility on the structure if you'd like. I guess we can move on to the integrating other issues? Interface residues or glycosylation sites are still open as far as I can see. Do you have any other ideas?

[mlgill]

@Ninjani I'm happy to work on adding the SWISS-MODEL annotation URL to my current PR, if you'd like. This code looks like a reference for how to do that, unless you have another example in mind?

After that I'll look into the glycosylation sites work.

[Ninjani]

Yes, I think you can use this script from the utils folder combined with the parse_pdbe.py script from that pull request to make the annotation file.

Okay, cool!

[mlgill]

Currently working on incorporating the solvent accessibility measurements into @Ninjani 's framework. Had several work interruptions today, but will finish debugging and testing tomorrow.

[Ninjani]

@mlgill Made some fixes to the parse_pdbe.py file (See #38). There were potentially some bugs in the residue mapping before as we were relying on alignment - now it uses mappings derived from PDBe so should be more correct.

[mlgill]

@Ninjani Could you verify that the branch I should be pulling code from in your fork is "master" and not "utils"? Thank you!

[gtauriello]

@all-contributors please add @akdel for code

[allcontributors]

@gtauriello

I've put up a pull request to add @akdel! 🎉

[gtauriello]

@all-contributors please add @mlgill for code

[allcontributors]

@gtauriello

I've put up a pull request to add @mlgill! 🎉