I am trying to understand the difference between the Rosetta and the POEM@home projects. In the POEM forum, I found a post that explains it like this:

As a quick first answer: Protein simulation is presently dominated by two approaches.

1) You learn/copy from nature (ROSETTA@HOME is a good example). Given a new sequence, one tries to copy structural fragments of known proteins and assembles them to good structures. Pros: Works for big proteins, gives excellent structures for high sequence similarity. Cons: Does not work, when there is no sequence similarity (new folds), does not work when there is no experimental data (transmembrane proteins, to which 40% of all known drugs bind), no kinetics

2) You simulate the folding process as it occurs in nature (Folding@Home) with a biophysical model. This is done with molecular dynamics (MD), an essentially sequential method with a time resolution of 10E-15 s/step. For a folding process in millisecond range you need A LOT of steps. Pros: full dynamics info, folding times, high accuracy structures, Cons: works only for small proteins, specific questions

POEM tries to interpolate between these two worlds. It uses an atomistic model for the protein free energy, i.e. is can work for new folds and applications in nanobiotechnology, where there is no experimental data. In contrast to MD, it exploits Anfinsons thermodynamic hypothesis (Nobel Prize in Chemistry 1972) that proteins in their biologically active state have a minimal free energy. The simulation process is thus replaced by an optimization process that is thousands of times faster than MD. Pros: Can do at least medium size proteins, gets the folding landscape, works for "new folds", Cons: still limited to proteins < 100 amino acids, no real kinetics (yet).

With POEM@HOME we will try to make progress on these two cons. Specifically we hope to make progress for


* "new fold" proteins with low sequence similarity to existing proteins
* proteins in nonphysiological environments (we just got a grant to develop implants, which are more biocompatible)
* understanding the folding process of more complex proteins that cannot be studied with direct kinetic simulation
* protein-protein interactions, where the partners change their conformation upon docking (biological signal process)
* refinement of model structures for transmembrane proteins

Is this a correct explanation, especially regarding Rosetta? The statement that Rosetta doesn't work for new folds surprised me a bit.

I am trying to understand the difference between the Rosetta and the POEM@home projects. In the POEM forum, I found a post that explains it like this:

As a quick first answer: Protein simulation is presently dominated by two approaches.

1) You learn/copy from nature (ROSETTA@HOME is a good example). Given a new sequence, one tries to copy structural fragments of known proteins and assembles them to good structures. Pros: Works for big proteins, gives excellent structures for high sequence similarity. Cons: Does not work, when there is no sequence similarity (new folds), does not work when there is no experimental data (transmembrane proteins, to which 40% of all known drugs bind), no kinetics

2) You simulate the folding process as it occurs in nature (Folding@Home) with a biophysical model. This is done with molecular dynamics (MD), an essentially sequential method with a time resolution of 10E-15 s/step. For a folding process in millisecond range you need A LOT of steps. Pros: full dynamics info, folding times, high accuracy structures, Cons: works only for small proteins, specific questions

POEM tries to interpolate between these two worlds. It uses an atomistic model for the protein free energy, i.e. is can work for new folds and applications in nanobiotechnology, where there is no experimental data. In contrast to MD, it exploits Anfinsons thermodynamic hypothesis (Nobel Prize in Chemistry 1972) that proteins in their biologically active state have a minimal free energy. The simulation process is thus replaced by an optimization process that is thousands of times faster than MD. Pros: Can do at least medium size proteins, gets the folding landscape, works for "new folds", Cons: still limited to proteins < 100 amino acids, no real kinetics (yet).

With POEM@HOME we will try to make progress on these two cons. Specifically we hope to make progress for


* "new fold" proteins with low sequence similarity to existing proteins
* proteins in nonphysiological environments (we just got a grant to develop implants, which are more biocompatible)
* understanding the folding process of more complex proteins that cannot be studied with direct kinetic simulation
* protein-protein interactions, where the partners change their conformation upon docking (biological signal process)
* refinement of model structures for transmembrane proteins

Is this a correct explanation, especially regarding Rosetta? The statement that Rosetta doesn't work for new folds surprised me a bit.

It is wrong. Rosetta uses a physically detailed all atom model, and has been unquestionably the best method to date for new folds, as well as for designing new proteins. the person who wrote that text is clearly not very well informed about the results of the CASP experiments or recent progress in protein design.

---

Thank you for the answer. I did find the expanation strange. Do you at rosetta know enough about POEM in order to be able to explain the difference between it and rosetta?

In response to a question on this thread, I tried to find out about the science behind POEM. I didn't find anything on the web beyond the POEM home page, which has no details. anybody know more?

---

In response to a question on this thread, I tried to find out about the science behind POEM. I didn't find anything on the web beyond the POEM home page, which has no details. anybody know more?

If still needed, this page (no link found on the POEM@home page?) seems to present a little more background.

In response to a question on this thread, I tried to find out about the science behind POEM. I didn't find anything on the web beyond the POEM home page, which has no details. anybody know more?

If still needed, this page (no link found on the POEM@home page?) seems to present a little more background.

**your url contained " " which was messing up the link, i cleaned that up**

...

If still needed, this page (no link found on the POEM@home page?)...

The links on the POEM entry page are bold black and thus not so easy to recognize as links.

The one you posted can be found here :

"POEM@HOME implements a novel approach to understand these aspects of protein structure"

Jumping in here a bit late, but the following paragraph from POEM's page has me a little puzzled.

POEM tries to interpolate between these two worlds. It uses an atomistic model for the protein free energy, i.e. is can work for new folds and applications in nanobiotechnology, where there is no experimental data. In contrast to MD, it exploits Anfinsons thermodynamic hypothesis (Nobel Prize in Chemistry 1972) that proteins in their biologically active state have a minimal free energy. The simulation process is thus replaced by an optimization process that is thousands of times faster than MD. Pros: Can do at least medium size proteins, gets the folding landscape, works for "new folds", Cons: still limited to proteins < 100 amino acids, no real kinetics (yet).

(Emphasis mine)

The part that I've highlighted is what has me confused, because my limited understanding of what we're doing here appears to rely on precisely this property.

---

Jumping in here a bit late, but the following paragraph from POEM's page has me a little puzzled.

POEM tries to interpolate between these two worlds. It uses an atomistic model for the protein free energy, i.e. is can work for new folds and applications in nanobiotechnology, where there is no experimental data. In contrast to MD, it exploits Anfinsons thermodynamic hypothesis (Nobel Prize in Chemistry 1972) that proteins in their biologically active state have a minimal free energy. The simulation process is thus replaced by an optimization process that is thousands of times faster than MD. Pros: Can do at least medium size proteins, gets the folding landscape, works for "new folds", Cons: still limited to proteins < 100 amino acids, no real kinetics (yet).

(Emphasis mine)

The part that I've highlighted is what has me confused, because my limited understanding of what we're doing here appears to rely on precisely this property.

me too...

---

Jumping in here a bit late, but the following paragraph from POEM's page has me a little puzzled.

POEM tries to interpolate between these two worlds. It uses an atomistic model for the protein free energy, i.e. is can work for new folds and applications in nanobiotechnology, where there is no experimental data. In contrast to MD, it exploits Anfinsons thermodynamic hypothesis (Nobel Prize in Chemistry 1972) that proteins in their biologically active state have a minimal free energy. The simulation process is thus replaced by an optimization process that is thousands of times faster than MD. Pros: Can do at least medium size proteins, gets the folding landscape, works for "new folds", Cons: still limited to proteins < 100 amino acids, no real kinetics (yet).

(Emphasis mine)

The part that I've highlighted is what has me confused, because my limited understanding of what we're doing here appears to rely on precisely this property.

me too...

try reading this wiki article http://en.wikipedia.org/wiki/Anfinsen%27s_dogma
also a NIH thread http://profiles.nlm.nih.gov/KK/Views/Exhibit/narrative/protein.html
perhaps this article in nature simplifies it better
http://www.nature.com/embor/journal/v7/n3/full/7400655.html

Jumping in here a bit late, but the following paragraph from POEM's page has me a little puzzled.

POEM tries to interpolate between these two worlds. It uses an atomistic model for the protein free energy, i.e. is can work for new folds and applications in nanobiotechnology, where there is no experimental data. In contrast to MD, it exploits Anfinsons thermodynamic hypothesis (Nobel Prize in Chemistry 1972) that proteins in their biologically active state have a minimal free energy. The simulation process is thus replaced by an optimization process that is thousands of times faster than MD. Pros: Can do at least medium size proteins, gets the folding landscape, works for "new folds", Cons: still limited to proteins < 100 amino acids, no real kinetics (yet).

(Emphasis mine)

The part that I've highlighted is what has me confused, because my limited understanding of what we're doing here appears to rely on precisely this property.

me too...

You are both right to be confused--searching for the lowest energy structure is of course at the core of what we are all engaged in with rosetta@home

---

Why is it confusing? If the conjecture that proteins will be in their lowest energy state when folded in their active configuration is true, then surely it is obvious that they will be attempting to acheive the same end when folding a residue sequence?

---

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

Why is it confusing? If the conjecture that proteins will be in their lowest energy state when folded in their active configuration is true, then surely it is obvious that they will be attempting to acheive the same end when folding a residue sequence?

It's the first line that's misleading:
"POEM tries to interpolate between these two worlds."

Where Rosetta and Folding are the two worlds, where in fact Rosetta and POEM are attempting the same goal. In the forum Rosetta is described as requiring fragments as a starting point, which is also incorrect (can't find the link atm)...

---

So the author of the "puff" did not have complete knowledge of the two projects, (which are quoted as examples of concepts). I, for one, do not find that totally suprising since they are not directly involved with them.

None of the projects spell out in black and white what they are doing, it tends to be gleaned and teased out over a, (often lengthy), period - and the other projects are supposed to have time to do that?

Remember what the front page text is suposed to do, give a really brief really simple introduction as to what the project is basically about. If it went into minute details, the casual visitors would never read it, and may never join.

---

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

Poem@home is about to release an open CL gpu client that will greatly improve their processing power.

I was wondering I anyone looked further into the similarities and differences between rosetta@home and poem@home.

---

Poem@home is about to release an open CL gpu client that will greatly improve their processing power.

I was wondering I anyone looked further into the similarities and differences between rosetta@home and poem@home.

I've been reving my GPUs all this time waiting to crunch. I hope it supports ATI cards.

---

I've been reving my GPUs all this time waiting to crunch. I hope it supports ATI cards.

OpenCL is hw-indipendent. And poem runs on gpu single-precision, so ATI is good!!

---

POEM has found that there are some differences between running on an AMD/ATI GPU and running on an Nvidia GPU.

http://boinc.fzk.de/poem/forum_thread.php?id=631#4840

Their GPU application now appears to be running properly on AMD/ATI GPUs, but not on Nvidia GPUs. Still trying, though, and they'd like volunteers to help test those tries.

BOINC does not appear to be ready to allow OpenCL workunits for any GPUs other than those from AMD/ATI and Nvidia. OpenCL is meant to be hardware-independent, but BOINC's interface to OpenCL GPUs definitely isn't.

Rosetta@Home has already mentioned that their minirosetta application is so serial in nature that a GPU version would not run much faster than the CPU version, and might actually run slower instead.

Good candidates of programs for conversion to a GPU version require large groups of steps that can be done in any order, or even all at once, since they do not affect each other.