Page 1 of 1

Questions about finished WUs:

Posted: Mon Jul 15, 2019 10:38 pm
by MeeLee
We fold for a company.
We spend money, time, hardware on this company.
Every day we download and upload a few megabytes in WUs to and back from a server.

And we don't know what happens after the WUs are sent back.
Only occasionally does the FAH website get updated with news, that's pretty vague about any successes on what specific WUs helped science.
But we don't know when our folding has caused a breakthrough in diseases, or when WUs are just discarded, overlooked, forgotten, or archived for later study.


1- Where do we find the actual data of the WUs?Is there a site that lists them all?
2- Can we even publicly view these WUs or their results, or do only scientists have access to them?
3- Or where can we find the results of our WUs (if a WU contained a trigger, or if it was just another WU that had no specific outcome)?
4- Which program 'decodes' these results into viewable, workable results (eg: graphs, an animated display (like the demo viewer), or some textual data,....)?
5- Who reviews all these WUs?How many scientists analyze this data? We know the main scientists working on this data, but is it forwarded to students, or what happens to them?
My GPUs run 24/7, multiple of them. I'm sure that a single, multi GPU server, is providing more data in a day, than a single scientist can analyze in a day.
And I'm only somewhere on the bottom of the list of active multiGPU folders (some of them have 10x faster servers than me), not to mention the thousands of folders that fold slower than me.


I'm hoping there can be more openness and transparency//accountability towards people folding.
More than just achieving points in a score system, it would benefit us, knowing more about successful hits (WUs that triggered results), and perhaps even a way to visualize this data somehow.

I'm hoping that FAH in the future will be able to give back more data on WUs that contributed to success, and if possible, usernames of people that worked on those WUs.
It would be far more accomplishing to see a table with names of users that helped find a cure, than to just see a general user score...

Re: Questions about finished WUs:

Posted: Mon Jul 15, 2019 11:28 pm
by Joe_H
MeeLee wrote:We fold for a company.
You start with a falsehood, folding is for a consortium of research labs run at various universities. F@h was started in Dr. Pande's lab at Stanford, other labs have joined since then. Many involved in running those labs are former grad students and post-docs who studied at the Pande Lab.

How much anyone spends on this is voluntary and up to them.

As for your questions about individual WU's, they by themselves do not "contain a trigger" or have an outcome other than successful completion or failure. It takes the thousands of WU's assembled through the various Gen's and and then put together with the different Runs to give statistical probabilities for the different conformations and folding paths to both properly folded proteins and misfiled ones.

As I understand it, there is ongoing discussions on how to better link research results presented on papers to various folding projects.

Others may weigh in on other points.

Re: Questions about finished WUs:

Posted: Mon Jul 15, 2019 11:38 pm
by MeeLee
Perhaps. To us, Folding at Home is an umbrella.
But I doubt the misunderstanding matters to the questions I asked.

That there will be more feedback is definitely something I'd think more folders would like to see.

Re: Questions about finished WUs:

Posted: Tue Jul 16, 2019 12:05 am
by JimboPalmer
MeeLee wrote:We fold for a company.
We spend money, time, hardware on this company.
Every day we download and upload a few megabytes in WUs to and back from a server.
[I am just a donor here, same as you. I interpreted your statement differently than Joe_H and am replying to that different interpretation]
If you mean that you are using your company's resources, be very sure your have written permission to do so. Folding is great, but not if it gets you fired.
I folded for years on 55 PCs at a mental health organization. (They were my client, I swear) But I had written permission to do so. http://www.regionone.org/index.html
So make sure your employer knows the Internet cost, the electrical costs and the computational costs. (Here in MS, it costs electricity to compute 24/7, but then it costs again to cool the building 24/7)

Are you hoping for concise reports that you can show your employer so they can see results from their costs?

Here is a recent news article https://www.mskcc.org/blog/computationa ... -movements

All the results of F@H are publicly available to anyone, although, to be honest, you really need to be in the medical field to be interested in the data.

Re: Questions about finished WUs:

Posted: Tue Jul 16, 2019 12:21 am
by rafwiewiora
3: You can't decompose any success into individual WUs, results come from whole trajectories (even if your particular WU accomplished a rare event transition, I'm not going to know it was a rare event, unless I also know that thousands of other WUs didn't accomplish that transition).

4/5 -- Students, like me, spend months analyzing them. One of the programs (of many open source projects) we use, e.g. for Markov state model building, is PyEMMA: http://emma-project.org/latest/. The field is working hard on automated ways to analyze these data, which remains a bottleneck.

All papers that use F@h data are listed here: https://foldingathome.org/papers-results/

1/2: For example, all the data I've finished analyzing so far made this recent paper (open access): https://elifesciences.org/articles/45403 // MSK blog post: https://www.mskcc.org/blog/computationa ... -movements

All the F@h data generated by you guys for it is available to download through the Open Science Framework: https://osf.io/2h6p4/ (places like this, free for academics, are quite a new thing -- we have some catching up on putting many datasets out there, but you can always request it from authors of a paper) -- this lists the project numbers which were used, you can find the user names of everyone who's contributed here: https://apps.foldingathome.org/contribu ... 1710,11713

Frankly we need a better way to put all this info together on a per project basis -- it's on the to-do list. We're also working on making all data we've ever collected (not all of which ends up in papers) available to download -- this is still a bit tricky, given that free places like the Open Science Framework can't accommodate all those TBs, and places like AWS come with restrictions to be affordable. One idea we're exploring is having it downloadable directly from the work servers.

Re: Questions about finished WUs:

Posted: Tue Jul 16, 2019 12:33 am
by rafwiewiora
I should add that if anyone would like to contribute to better transparency by helping us with writing blog posts etc. then that could help us improve massively -- I'd totally love to write a weekly blog post about how my data collection and analysis is going, but there's simply no time. Both for this, and for faster upgrades of things like the cores, we don't have enough people, because we don't have money to pay them. Grant funders pay us to produce papers, for some maintenance and hardware upgrades -- all those other things that are nice to have, but not vital is essentially unpaid work whenever you can find any time to do it.

Re: Questions about finished WUs:

Posted: Tue Jul 16, 2019 9:28 am
by MeeLee
Interesting!
While I'm not a scientist, and not really interested in studying the diseases themselves, I do feel it's important for people folding to have some sort of tangible feedback from FAH.
I'm mainly thinking about questions my family asks "what is it you're doing? Why are you 'wasting' your money on that? Can you show us what it is you're doing?
Fact is, I can't even show them the viewer, because it's just showing a protein demo going back and forth. It's not even the protein or project my PC is working on, and not even I understand what it is I'm looking at, other than seeing what appears to be a cluster of balls and sticks, which I presume to be atoms. I hope so, because that's what I'm telling them...

Neither can I show them what projects I worked on, or what these projects have accomplished so far.
All I can show them is a number. A scoreboard, and how many 'wus' I've processed (which is meaningless, when short WUs can take 1 minute to complete, and long ones a full day), and the bill it costs to keep it all running
They don't understand. The only thing they do understand, is that I'm paying the bill.

I get that the staff is shorthanded. But I think publicity is also very important, and the viewer helped some with that.
Articles would help. Especially major breakthroughs.
Doesn't have to be a weekly thing, but even a (semi) annual summary of the findings this year, will come a long way.

Even if we don't get our name on the site, at least I'd expect to see the most important discoveries and breakthroughs on the news site of FAH, together with a small breakdowns of what's found, what it means to a common man (not only scientific jargon).

I don't expect access to all WUs, but just something that can be presented to my family and friends, and to the world, of what's going on behind the scenes.
Eg: an example of a cancer cell forming, with an animated gif, and a small explanation of what we're seeing.
If people can understand it better, it would also make my life easier to get them aboard.
Most see something, and are in awe for what I'm doing, but it's not enough to convince them to go through the trouble of figuring out how to fold... Or even leave their PCs on overnight.
And most of them didn't go through the trouble of learning Linux, and are just running Windows, so their learning curve shouldn't be that steep at all!

Re: Questions about finished WUs:

Posted: Wed Jul 17, 2019 12:40 am
by bruce
@MeeLee
FAH is all about science. As a general rule, Scientists don't do publicity (and/or they don't do it well). Their job should be focused on scientific matters that need to be shared with other scientists. Somebody else generally needs to rewrite their findings into layman's language. A magazine like Scientific American generally publishes the science so that well-educated non-scientists and understand what's going on. Then other magazines turn that article into popular science, that less well-educated folks can understand a smaller portion of what was discovered.

The articles you're seeking have an audience between the last two groups -- and it won't be published in the blog unless somebody writes it.

You, however, need to do your part to help them out. Start from the list of papers that have been published and read them. They are generally written by scientists for an audience of scientists, so you'll probably have trouble understanding them, but one or two of them will have a section that you do understand. Rewrite that part of that paper so that your family/friends can understand what's being said in the original paper.

What you write will be helpful to your peers here if it's published in FAH's blog, but it won't be ready for that to happen. It can be submitted to some of FAH experts for a sort of peer-review ... and it may end up being published in the FAH blog, too after being edited by others --- but not as a scientific paper. The objective is for you to help fulfill FAH's need to keep all of us (the Donors, not the Scientists) informed -- just as you said that "somebody" needs to do that. ... and by having a scientist review it, it can be a preliminary suggestion for a blog post.

Re: Questions about finished WUs:

Posted: Wed Jul 17, 2019 6:13 pm
by rafwiewiora
Just to add on the papers being written in scientific language - some journals do layman descriptions now -- the paper I linked to above for example has an 'eLife digest'.

Re: Questions about finished WUs:

Posted: Thu Jul 18, 2019 1:25 am
by bruce
I think that the point is that articles posted in FAH's "blog" are becoming very rare now that VIjay isn't publishing anything. Posting the layman's version when papers are accepted for publication would be a big help toward keeping the Donors seeing fresh information periodically-- and let them know how their completed WUs contribute to research.