Hey Jeff! I think qualifiers can help clear up the usage of the term "ensemble":

A representative ensemble could be a collection of models that are highly probable (according to the modeling process being used) and have been filtered from a larger set of models. This is probably what a depositor would submit to the PDB for archiving, but is ultimately somewhat arbitrary.

The usage of statistical ensemble and thermodynamic ensemble have clear meanings within the physics literature, so those should be safe.

In the case that an ensemble is meant to refer to a representation of a posterior distribution, like what MCMC produces, then I think the standard term sample should be preferred, since a sample can include improbable (low scoring) models, but is guaranteed to do so in proportion to the posterior density.

Thanks for looking into Slack! Here's a list of names/e-mails:

Aji Palar aji@salilab.org
Arthur Zalevsky aozalevsky@gmail.com
Andrew Latham aplatham@salilab.org
Brinda Vallat brinda.vallat@rcsb.org
Tracy Wang chenxi.wang@ucsf.edu
Matthew Hancock matthewhancock97@gmail.com
Neelesh Soni neeleshsoni03@gmail.com

If anyone prefers a different e-mail than what I have, I suppose they can reach out to you.

On Wed, Jun 28, 2023 at 3:06 PM Hoch,Jeffrey <hoch@uchc.edu> wrote:

Thanks Jared.

 

I think the terms so far should work – however we should make sure our terms for ensembles align with thermodynamic ensembles (microcanonical, canonical, grand canonical...others).

 

Also we need terms to describe the scope of averaging for forward modeling – which may or may not cover all of or align with an ensemble.

 

We looked into a paid-up Slack license. A bit over $10pp/yr . I think I can handle this. Can you assemble a list of names and email addresses of those who should be included? I’ll have Jen Reeser set up the account.

 

Yours, Jeff

 

From: Jared Sagendorf <jared.sagendorf@rcsb.org>
Date: Friday, June 23, 2023 at 5:57 PM
To: "Hoch,Jeffrey" <hoch@uchc.edu>, "bayesian_validation@salilab.org" <bayesian_validation@salilab.org>
Cc: "Pustovalova,Yulia" <ypustovalova@uchc.edu>, "Eghbalnia,Hamid R." <heghbalnia@uchc.edu>, "Pozhidaeva,Alexandra" <pozhidaeva@uchc.edu>, "Courtney,Joseph M." <jcourtney@uchc.edu>, "Gryk,Michael R." <gryk@uchc.edu>, "Baskaran,Kumaran" <baskaran@uchc.edu>
Subject: Re: Next meeting agenda

 

*** Attention: This is an external email. Use caution responding, opening attachments or clicking on links. ***

Jeff will look into getting a Slack set up with a premium subscription and we can continue refining the terminology as we go.

I have a shared google drive set up through the RCSB google organization, but unfortunately I can't seem to invite people with non-gmail e-mails - if anyone knows a workaround let me know!

- Jared

 

 

 

On Wed, Jun 21, 2023 at 3:26 PM Jared Sagendorf <jared.sagendorf@rcsb.org> wrote:

Sure Jeff, pushing it forward two hours to 12:00 PST should work for most of us. Hopefully that makes it easier for you! I'll update the calendar event on our side.

Congrats on the R24!

See you Friday,

 

- Jared

 

On Wed, Jun 21, 2023 at 12:20 PM Hoch,Jeffrey <hoch@uchc.edu> wrote:

Hi Jared – met too! I added a term to the dictionary 😉. Yulia, Michael and I will be driving back from a Gordon Conference in Vermont Friday morning, and it might be tight for us to make the schedule time. I know there are many schedules to check, but would it be possible to bump the time back a couple of hours to take some of the stress out? -Jeff

 

PS – good news is that we heard BMRB will be funded via an R24 mechanism. Don’t know the budget yet, though.

 

From: Jared Sagendorf <jared.sagendorf@rcsb.org>
Date: Wednesday, June 21, 2023 at 2:08 PM
To: "Hoch,Jeffrey" <hoch@uchc.edu>
Subject: Next meeting agenda

 

*** Attention: This is an external email. Use caution responding, opening attachments or clicking on links. ***

Hi Jeff, looking forward to the next meeting!

For the agenda this week, I was thinking we could start with having a look at some of the terminology, now that we have a dictionary in place. I'll put together a few slides that I hope will explain the way the Sali group speaks about modeling, and we can identify if there's any inconsistencies between the two groups.

Aside from that, I thought it would be good to begin a discussion on what the current hurdles are to actually implementing a fully Bayesian validation pipeline for the PDB. What are the unsolved problems or unimplemented steps that are missing to actually push this forward, and could we work together to start solving those issues?

Also happy to send out any items you would like to bring up!

 

- Jared