| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: OpenPandemics - COVID-19 Project
Thread: "September 2020 Update: Fight AIDS@Home" article is relevant to OpenPandemics |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 15
|
|
| Author |
|
|
Rickjb
Veteran Cruncher Australia Joined: Sep 17, 2006 Post Count: 666 Status: Offline Project Badges: 
|
The WCG News Section article September (2020) Update: Fight AIDS@Home discusses the future of the Fight AIDS@Home project (FAAH), which is currently paused.
----------------------------------------OpenPandemics is doing basically the same things as FAAH - trying to predict the strength of attraction between drug potential candidate molecules (ligands) and critical molecules in the pathogens. Thus some of the statements in the FAAH update are relevant to OPN: "... in the past few weeks (the FAAH research teams') discussion has turned to how to best use new approaches and technologies to further refine and possibly accelerate their data analyses. They are now in the very early stages of exploring what new approaches and technologies (if any) may help streamline their work." I suggest that the bit about technologies and streamlining refers to using GPUs. "Further refining" may refer to improving the accuracy of the modelling, in whcih case I sure hope they can find ways forward. Three different modelling programs - AutoDock, VINA and BEDAM - have been used for FAAH, and AFAIK they gave 3 different sets of predictions. At most one of them can be giving accurate predictions. AutoDock and VINA both give too many false positives on proteins from several viruses. It is likely that AutoDock will also be giving false positive predictions for COVID-19, but finding better treatments for infected people is urgent at this time, and hopefully our best strategy is the carry on with AutoDock for now and filter out false positives in some way later ... [Edit 1 times, last edit by Rickjb at Sep 17, 2020 4:22:35 PM] |
||
|
|
Falconet
Master Cruncher Portugal Joined: Mar 9, 2009 Post Count: 3315 Status: Offline Project Badges: 
|
I'm very eager to learn the results from FAHB data we finished a couple of months ago. Hope there are a couple of good candidates from those 21 or so compounds.
----------------------------------------Accuracy. It is what it is. Hopefully there is room for improvement. The Rosetta people created over 2 million protein designs to target SARS-CoV-2 and ran them on Rosetta@home and other computing resources they had access to (in-house, a Microsoft Cloud offer, etc). They tested in the lab over 20,000 of the most promising. I think I recall reading only a couple hundred actually worked well (since they don't post news or otherwise participate over at Rosetta@home, I'm remembering this from memory). So, out of the 20,000 most promising "minibinders" from the computing simulations, only a few worked. Anyway, now they have a vaccine candidate they say is better than other vaccines and also antiviral drug candidates, all of which they are working to get to human trials. https://www.ipd.uw.edu/wp-content/uploads/2020/08/IPD_AnnualReport_2020-web.pdf Any improvement is welcomed, but many false positives are something to be expected, I suppose. That said, even with false positives, AutoDock and Vina work and it's those 2 or 3 good hits on the lab that we want from the result of our AutoDock computations. GPUs, hope we get an update on the next monthly OPN update.  - AMD Ryzen 5 1600AF 6C/12T 3.2 GHz - 85W - AMD Ryzen 5 2500U 4C/8T 2.0 GHz - 28W - AMD Ryzen 7 7730U 8C/16T 3.0 GHz [Edit 1 times, last edit by Mosqueteiro at Sep 15, 2020 4:03:34 PM] |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges: 
|
"Further refining" may refer to improving the accuracy of the modelling, in whcih case I sure hope they can find ways forward. Three different modelling programs - AutoDock, VINA and BEDAM - have been used for FAAH, and AFAIK they gave 3 different sets of predictions. At most one of them can be giving accurate predictions. AutoDock and VINA both give too many false positives on proteins from several viruses. That is why FAAH has a Phase 1 and a Phase 2. Phase 1 runs on AutoDock/VINA and performs a first screen. The most promising ones are then sent to Phase 2, where they are screened by BEDAM, which is slower but more accurate. If they could use the GPUs for a fast Phase 1 on OPN, they could then forward the results to the CPUs for a Phase 2. But they have not said that yet, only that they are trying to get the GPUs to run the same work units as the CPUs. |
||
|
|
Falconet
Master Cruncher Portugal Joined: Mar 9, 2009 Post Count: 3315 Status: Offline Project Badges:
|
"Further refining" may refer to improving the accuracy of the modelling, in whcih case I sure hope they can find ways forward. Three different modelling programs - AutoDock, VINA and BEDAM - have been used for FAAH, and AFAIK they gave 3 different sets of predictions. At most one of them can be giving accurate predictions. AutoDock and VINA both give too many false positives on proteins from several viruses. That is why FAAH has a Phase 1 and a Phase 2. Phase 1 runs on AutoDock/VINA and performs a first screen. The most promising ones are then sent to Phase 2, where they are screened by BEDAM, which is slower but more accurate. If they could use the GPUs for a fast Phase 1 on OPN, they could then forward the results to the CPUs for a Phase 2. But they have not said that yet, only that they are trying to get the GPUs to run the same work units as the CPUs. The Overview page for FAH2 has a good summary on AutoDock false positives and the use of BEDAM. - AMD Ryzen 5 1600AF 6C/12T 3.2 GHz - 85W - AMD Ryzen 5 2500U 4C/8T 2.0 GHz - 28W - AMD Ryzen 7 7730U 8C/16T 3.0 GHz |
||
|
|
Aurum
Master Cruncher The Great Basin Joined: Dec 24, 2017 Post Count: 2391 Status: Offline Project Badges: 
|
OpenPandemics is doing basically the same things as FAAH... You're comparing apples and oranges. HIV is a retrovirus and SARS is an RNA virus. Dennis Burton explained it in one of the Scripps videos."... in the past few weeks (the FAAH research teams') discussion has turned to how to best use new approaches and technologies to further refine and possibly accelerate their data analyses. They are now in the very early stages of exploring what new approaches and technologies (if any) may help streamline their work."   ...KRI please cancel all shadow-banning |
||
|
|
yoerik
Senior Cruncher Canada Joined: Mar 24, 2020 Post Count: 413 Status: Offline Project Badges:
|
The point of a project like this is to narrow the possibilities down from infinite, to a tangible number. That just makes sense -- it's the job we're doing on projects like this.
---------------------------------------- |
||
|
|
Aurum
Master Cruncher The Great Basin Joined: Dec 24, 2017 Post Count: 2391 Status: Offline Project Badges:
|
The Rosetta people created over 2 million protein designs to target SARS-CoV-2 and ran them on Rosetta@home and other computing resources they had access to (in-house, a Microsoft Cloud offer, etc). They tested in the lab over 20,000 of the most promising. I think I recall reading only a couple hundred actually worked well (since they don't post news or otherwise participate over at Rosetta@home, I'm remembering this from memory). So, out of the 20,000 most promising "minibinders" from the computing simulations, only a few worked. Might this be the paper you're referring to? https://science.sciencemag.org/content/early/2020/09/08/science.abd9909...KRI please cancel all shadow-banning |
||
|
|
Aurum
Master Cruncher The Great Basin Joined: Dec 24, 2017 Post Count: 2391 Status: Offline Project Badges:
|
If they could use the GPUs for a fast Phase 1 on OPN, they could then forward the results to the CPUs for a Phase 2. And yet IBM wants to shove all OPN WUs down the same pipe so how will that be possible???But they have not said that yet, only that they are trying to get the GPUs to run the same work units as the CPUs. ...KRI please cancel all shadow-banning |
||
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges: 
|
If they could use the GPUs for a fast Phase 1 on OPN, they could then forward the results to the CPUs for a Phase 2. And yet IBM wants to shove all OPN WUs down the same pipe so how will that be possible???But they have not said that yet, only that they are trying to get the GPUs to run the same work units as the CPUs. Isn't that call made by the research team? |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges:
|
I know of several projects that use both CPUs and GPUs on the same OpenCl work units. At the moment, Einstein does on the gravity wave work units, and MilkyWay is planning to do so shortly on the N-body work. In the past, there was also POEM.
I am all for using the GPUs on CUDA work units if they can, but as BladeD says, they are the scientists. You would have to have a really detailed knowledge of what they are doing to really suggest anything worthwhile. |
||
|
|
|