This thread is the fourth of a series where participants can discuss and ask questions about Dr. Baker's journal entries.

To reference discussions prior to this, see Discussion 3.

---

Rosetta Informational Moderator: Mod.Zilla

Can you make a similar comparison for CASP8 as POEM did here?

I suppose you have numbers, something like probability that predicted model is the real model. 2-3 years ago it was close to zero, what percentage is real now? and if it's possible, can you show dynamics for the last few years?

These are of course excellent questions. we will have full answers in December at the CASP meeting. because not all target structures have been released, it is not possible now to give definitive answers. the probability of predicting the native structure with high accuracy is increaseing, but is still pretty small--this is why with rosetta@home we are still focusing primarily on developing and testing improved methods.

---

Can you make a similar comparison for CASP8 as POEM did here?

Stuff like this posted on homepage, or on a section linked from homepage makes the Project seem alive and prosperous.

Excellent idea.
Plus, it would probably reduce the amount of people that "drop-out" off the project because they don't see any team participation. (Most of team participation happens within the forum)

---

Can you make a similar comparison for CASP8 as POEM did here?

Data for such comparisons is available through several preliminary CASP8 assessments available here: http://www.reading.ac.uk/bioinf/CASP8/.

It would be even more interesting to see some superpositions (like this one of TOP7) of a few high-scoring Rosetta@home structure predictions onto their solved structures (available in the PDB). I described a way to get Rosetta's CASP8 sutrcture predictions here: https://boinc.bakerlab.org/rosetta/forum_thread.php?id=4371&nowrap=true#55719. To superposition proteins in PyMOL, I think the syntax is something like "align nameOfStructure1, nameOfStructure2".

---

Can you make a similar comparison for CASP8 as POEM did here?

Data for such comparisons is available through several preliminary CASP8 assessments available here: http://www.reading.ac.uk/bioinf/CASP8/.

It would be even more interesting to see some superpositions (like this one of TOP7) of a few high-scoring Rosetta@home structure predictions onto their solved structures (available in the PDB). I described a way to get Rosetta's CASP8 sutrcture predictions here: https://boinc.bakerlab.org/rosetta/forum_thread.php?id=4371&nowrap=true#55719. To superposition proteins in PyMOL, I think the syntax is something like "align nameOfStructure1, nameOfStructure2".

One thing to keep in mind is that all the comparisons so far are between the automatic servers as the "human" group results are not yet public. (this is why there is a "ROBETTA" entry, but not a "ROSETTA" entry--almost all the rosetta@home work this summer was for the human group not the robetta server). if you look at the sites mentioned above, try to guess where your rosetta@home results rank on the list!

---

If you split your computational resources between several different projects this would be a good time to temporarily increase rosetta@home's share.

That raised my eyebrow. I think it inappropriate for project managers to ask people to move crunching power away from other projects to their own.

---

Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.

If you split your computational resources between several different projects this would be a good time to temporarily increase rosetta@home's share.

That raised my eyebrow. I think it inappropriate for project managers to ask people to move crunching power away from other projects to their own.

Projects often ask for more power, mostly during competitions or special projects. Nothing wrong with asking, IMO. It's up to the crunchers to decide.

---

If you split your computational resources between several different projects this would be a good time to temporarily increase rosetta@home's share.

That raised my eyebrow. I think it inappropriate for project managers to ask people to move crunching power away from other projects to their own.

Projects often ask for more power, mostly during competitions or special projects. Nothing wrong with asking, IMO. It's up to the crunchers to decide.

I second that view

When there are competitions, it is usually, in my experience, teams and users that are pushing. I crunch a lot of projects and am active on many of the projects forums, but it is not my experience that project managers OFTEN ask for more.

---

Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.

When there are competitions, it is usually, in my experience, teams and users that are pushing. I crunch a lot of projects and am active on many of the projects forums, but it is not my experience that project managers OFTEN ask for more.

Even if Rosetta@home was the only project doing it, I still don't think it would be inappropriate. No harm in asking.

What would be inappropriate is if the project somehow had control over our BOINC resources sharing ratio and changed it remotely without asking, or something like that.

---

When there are competitions, it is usually, in my experience, teams and users that are pushing. I crunch a lot of projects and am active on many of the projects forums, but it is not my experience that project managers OFTEN ask for more.

Even if Rosetta@home was the only project doing it, I still don't think it would be inappropriate. No harm in asking.

What would be inappropriate is if the project somehow had control over our BOINC resources sharing ratio and changed it remotely without asking, or something like that.

You said that.

When there are competitions, it is usually, in my experience, teams and users that are pushing. I crunch a lot of projects and am active on many of the projects forums, but it is not my experience that project managers OFTEN ask for more.

Even if Rosetta@home was the only project doing it, I still don't think it would be inappropriate. No harm in asking.

What would be inappropriate is if the project somehow had control over our BOINC resources sharing ratio and changed it remotely without asking, or something like that.

well I learned my lesson in any event--sorry about that!

---

David, no harm in asking at all and you should ask again. If you cannot trumpet your own project in your own forums what can you do, lol.

Only here as I've just been reading the last few Journal entries (only thread I subscribe to and really should be a main page and RSS fed column by the way).

since it's the most informative and useful thread on the site (and it's buried away in the forum).

---

Team mauisun.org

If I'm reading the results correctly (its not the easiest table to interpret and there's a ton of information in there), it seems that R@h did better overall in CASP8 than any of the other 164 participants. That's wonderful!

---

The CASP8 summary stats links is hard to understand, can someone do a writeup that explains how Rosetta did in comparison to everyone else, anything exciting that happened, stuff like that?

---

It seems like they're on to something pretty big in that last post.

---

If I'm reading the results correctly (its not the easiest table to interpret and there's a ton of information in there), it seems that R@h did better overall in CASP8 than any of the other 164 participants. That's wonderful!

That's also what I make of it. And if I'm right, we don't just win in a close call, we win hands down!
If this true, it should in my opinion be front page news! This is fantastic!

If I'm reading the results correctly (its not the easiest table to interpret and there's a ton of information in there), it seems that R@h did better overall in CASP8 than any of the other 164 participants. That's wonderful!

That's also what I make of it. And if I'm right, we don't just win in a close call, we win hands down!
If this true, it should in my opinion be front page news! This is fantastic!

Yes, I'd also have thought they could have featured this a little more prominently in the news section, if only to provide a bit of positive reinforcement for all those who contribute computer time to this project.

Its also worth checking out http://fold.it/portal/blog where, among other things, the Rosetta developers discuss the (very good) CASP8 performance of humans playing FoldIt, the interactive version of Rosetta@home

---