 |
What is Peer or Peer Computing? |
 |
Sometimes known as Distributed Computing, Peer to Peer Computing harnesses the power
of a network of computers to perform some computations. For Find-a-Drug and
similar internet projects, the computational programs run in background or as a screen-saver
application so that they do not significantly intrude on the normal use of the computer.
|
|
Can any project be adapted for Peer to Peer Computing? |
|
No. The project has to be divided into a large number of independent calculations
which can be distributed to members computers and the results returned to the server(s).
|
|
What other aspects are important for such projects? |
|
There are practical restrictions concerning the amount of data that may
be distributed and the communication frequency with the server(s). This in
turn impacts the choice of algorithms and the number of members a server can support.
|
|
How many servers are used by Find-a-Drug? |
|
Unlike some similar projects, we are currently using six in different locations and envisage this number to
grow as the number of members also grows. This avoids contention for access to the server
as well as redundancy in the event of a network or server failure.
|
|
Where can I learn more about distributed computing? |
|
Two relevant web-sites are Internet-based Distributed Computing Projects and
Intel Philanthropic peer-to-peer program both of which also provide information about other projects..
|
|
What are statistics?
|
|
Members contributions are reported and analysed by team and country in the
stats.
We report the progress of the projects in the project stats pages.
These pages analyse for each target the numbers of jobs processed and the corresponding number of
molecules, hits and time taken.
|
|
What are teams?
|
|
Teams are groups of members for whom the numbers of molecules and hits are totalled. Click
here for more information about the statistical reports on the
calculations and what teams exist.
|
|
How can I form my own team?
|
|
Each team must have a web-site, a team captain and a name. You need to e-mail
this information to teams@find-a-drug.org.uk,
who will assign a team ID. If your web-site is a single page plus reasonable graphics,
we will host it. Otherwise, it can be hosted by several other organisation free
of charge (eg www.freeserve.com). You may also provide an
team icon which is displayed on your team stats page.
|
|
What is the best measure of a member's contribution?
|
|
The time taken per job does vary and in some cases the number of molecules processed
is significantly different. Usually jobs which give a large number of hits take
significantly longer than those which give only a few. The number of Points or GFlops is usually a
better indication.
|
|
What are GFLOPS?
|
|
GFLOPS is a measure of the processing performed. It is an abbreviation for Giga Floating point
OPerationS. This supercedes our POINTS system which was a more abstract measure of performance.
For an average CPU (1.2GHz) this will result in approximately 800 GFLOPS per CPU hour. (Each Point
corresponds to 8 GFLOPS).
|
|
What precautions are taken to ensure the security of information sent from my PC?
|
|
All information uploaded to Find-a-Drug servers is encrypted and is
transferred to computers inaccessible to the public prior to de-encrypting.
|
|
Is it conceivable that Find-a-Drug software introduces a virus or
other harmful agent on to my computer?
|
|
In order to do so it would be necessary for some-one to hack undetected into
Treweren Consultant's computers or Find-a-Drug servers. We take the
usual precautions to prevent this and recommend that members continue
to run virus detection software. In addition, the download files
are regularly replaced limiting the consequences of any infection.
|
|
What information does Find-a-Drug upload from my PC?
|
|
The only information sent to our servers are the calculations results,
time and configuration information which you have entered in the Find-a-Drug
control panel. We perform no checks on the accuracy of the processor
information, country, e-mail address etc.
|
|
What happens if I provide am invalid or incorrect e-mail address?
|
|
The e-mail address is used to send quarterly newsletters and if you
participate in the beta project information about developments. If you
join a team you may elect to receive e-mails from the team captain. If
your e-mail address is not valid, we have no means of contacting you.
|
|
Who else has access to the information I provide?
|
|
The information entered in the Find-a-Drug control panel is encrypted
prior to being sent to a Find-a-Drug server and is only de-encrypted
on Treweren Consultant's computers which cannot be accessed externally.
The information is not and will not be disclosed to third parties.
|
|
What are the most common connection problems?
|
|
- Firewall configuration preventing fadsetup and server accessing the Internet
- Internet Explorer configured for off-line use
- Proxy configuration issues
|
|
Will a fire-wall prevent the Find-a-Drug control panel (fadsetup) and think working normally?
|
|
Perhaps. The Find-a-Drug control panel (fadsetup) and the job queue manager (server)
use the HTTP to communicate with one of a list of Find-a-Drug servers.
You may need to change you fire-wall setup to enable these programs
to send and receive data from the servers listed in http.lis and cgi.lis.
|
|
The screen saver is reporting that THINK is idle or showing only the Mortar Board hat.
|
|
This indicates that more jobs need to be downloaded from a Find-a-Drug server and is normal
when the software is first installed for up to 2 minutes.
|
|
Despite being idle, no connections are made to the Find-a-Drug server
|
|
Check that the server process is running. If not, logout and login and then
check again. If this continues to be the case then check that a file named think.lgo
is located in an appropriate startup folder and that the filetype lgo is registered
to start the THINK loader application (using Windows Explorer click on Folder Options on
the Tools menu and then File Types).
|
|
The server process is running and attempting to connect to the server,
but THINK remains idle |
|
If there are no jobs for any of the projects you have selected this behaviour is to be
expected. Check for messages in server.log.
|
|
The error message "(10522) Connection failure" is being returned by the Job Queue Server?
|
|
When this is being reported during registration and/or the first time the
Job Queue Server tries to download jobs, the usual cause is a firewall configuration
which prevents fadsetup.exe and/or server.exe access the Find-a-Drug
Internet servers. Obviously, if you change the firewall setup later this
problem can re-occur. The URLs of the servers are listed in http.lis and cgi.lis.
We have also observed that on some occasions Windows can remember that a
server was unavailable when it last tried and give an error without re-trying!
Under these circumstances the problem clears if a dial-up Interconnection is
restarted, the Job Queue Server restarted or the PC rebooted.
|
|
Do you currently support connections through a proxy server?
|
|
Under windows the software automatically uses the proxy settings stored for Internet Explorer.
Under Linux you need to insert the following environment variables by editing think.env
| THINK_PROXY_HOST=proxy_server |
| THINK_PROXY_PORT=port |
| THINK_PROXY_USER=user |
| THINK_PROXY_PASSWORD=password |
The default port is 8080 and is the user or password is unspecified then it is assumed to be blank.
|
|
How frequently does THINK connect to the internet?
|
|
You can control the frequency of planned connections in the Find-a-Drug
control panel. If THINK completes all queued jobs, then an unplanned connection will
occur if there are no molecules files from previous protein targets which can be
re-used on the current target. If you select "When On-line" then no attempt
to dial-up will be made. The "None" setting should be used when you wish to manually download
jobs and upload results.
|
|
How long does THINK take to complete a job?
|
|
This depends on the speed of your computer and how fast it processes each
set of molecules against the query. Most average jobs take a few hours on an average PC.
The project stats pages provide
some analysis of the time taken, number of hits, redundancy etc for each protein target.
|
|
Can I disable automatic dial-up to connect to the internet?
|
|
Yes, use the Find-a-Drug control panel (fadsetup.exe) and set Planned Connections
to "After Confirmation". You can also configure the software to only
upload results and download new jobs when the computer is already on-line by
using the "When on-line" option under Planned Connections.
|
|
I have enabled automatic dial-up but a connection dialog continues to appear
|
|
This is the default Windows connection dialog unless your ISP has provided
a replacement. The default Windows dialog (for most versions) includes a toggle
"Connect Automatically" but for ISP dialogs this is often hidden as a sub-dialog.
If you can't find it from the connection dialog, try other dialup networking dialogs.
|
|
My PC does not disconnect after connecting to the Find-a-Drug server
|
|
The normal Windows dialogs includes a sub-dialog with a "disconnection
when idle" option. The exact details of this vary although some include a
"low traffic" disconnect rather than "zero traffic" disconnect.
This is preferrable because there have been reports of other software polling
the internet and retaining the connection. It is believed that the AOL custom
dialogs do not include an automatic disconnect feature!
|
|
My PC is disconnecting when I leave Outlook Express and prematurely terminating the
download from Find-a-Drug
|
|
There is an option in Outlook Express (Tools|Options|Connections) to disconnect after
sending and receiving e-mails. Disable this option!
|
|
How much data is transferred?
|
|
This does vary. The minimum downloaded is about 5Kbytes, but when a new protein
target is required a further 50Kbytes will be transferred. These quantities are small
compared to most other Distributing Computing projects are will normally be transferred
in much less than 1 minute.
|
|
How do you select projects?
|
|
The most important aspects of a project are to ascertain the scientific feasibility
of the project and the value to society.
|
|
Who funds Find-a-Drug?
|
|
At present, Find-a-Drug is funded by Treweren Consultants (the developers of
the THINK software). In the future, we expect some revenue from partners who
will fund some of the research and royalties from discoveries.
|
|
Who benefits from the projects?
|
|
Find-a-Drug operates on a not for profit basis. This means that
all revenues (including any royalties from the sales of any resulting drugs)
in excess of costs will be donated to research.
|
|
Who controls what happens to the results?
|
|
When there is an external partner, the partner is responsible for deciding
which molecules to make and test.
|
|
Can I opt in or out of projects?
|
|
Yes, just run the Find-a-Drug control panel (fadsetup.exe) and change the toggles.
|
|
Where can I find out more about the projects?
|
|
Click here for the Projects pages.
|
|
What does the software do?
|
|
The Find-a-Drug software downloads a job consisting of a protein target and
a set of 100 molecules from the server. When your PC is otherwise idle, it generates
about 100 derivatives for each molecule giving a total of 10,100 molecules. The
software then generates all the possible shapes or conformations of each molecule
and determines whether it can interact favourably with the receptor in the protein
target.
|
|
What can I see on the screen?
|
|
THINK displays the conformation of the molecule it is currently processing
and the progress of the calculation through the job. Faster PCs download several
jobs at the same time so that the "percentage complete" will rise to
100 and then fall to 0 without connecting to the internet.
|
|
What is the significance of the colours of the balls and sticks?
|
|
Each ball corresponds to an atom and each stick to a chemical bond between two atoms
and these are coloured in accordance
with the elements. The common colours are: blue for nitrogen,
red for oxygen, yellow for sulphur and cyan for carbon.
|
|
What is the significance of the Mortar Board or academic hat?
|
|
This is the THINK logo and it is displayed when THINK is not generating conformers
for a molecule - THINK may be generating derivatives or waiting for data.
|
|
How many conformers (or conformations) does a molecule have?
|
|
This varies from tens to potentially millions of shapes. However, these are grouped into similar
sets in such a way that when one conformer is rejected many more may also be rejected at
the same time as the program progresses to a different shape.
|
|
What is a hit?
|
|
This is a molecule which is predicted to interact or bind with the protein receptor.
|
|
How do you distinguish a good hit from a poor hit?
|
|
The hits are scored in such a way that those which score less than 0 are
predicted to bind to the protein receptor. Good inhibitors often have a score
less than -50 whereas molecules with scores above -25 can only be expected to
bind weakly.
|
|
How many molecules are there in each job?
|
|
This may change for different protein targets, but normally each job considers
10,100 molecules - 100 are downloaded from the server and 100 derivatives of each
molecule are then processed.
|
|
Where do the molecules come from?
|
|
They have been collected from public domain catalogues and combinatorial chemistry libraries of
molecules which are either commercially available or readily synthesised.
|
|
What is the difference between the version of THINK used for this project and that hosted by United Devices?
|
|
We are using the latest release of THINK which is substantially faster (in
part as a result of assistance provided by Intel) and refines all the crude hits
at the same time. This has the consequence of lowering the apparent hit rate because
only the best hits are returned to the server. In practice, this has an added
benefit: the internet data traffic is significantly reduced.
|
|
Where can I find out more information about THINK?
|
|
THINK is developed and maintained by Treweren
Consultants
|
|
Who owns the IPR in the results?
|
|
Members waive any Intellectual Property Rights (IPR) in the results in the licence agreement.
The computational predictions are confirmed by experiments by Find-a-Drug in partnership with
academics or small pharmaceutical companies. A fee is payable to Find-a-Drug to transfer
or licence the IPR. This fee may be paid on reaching certain milestones in
the drug development process or can take the form of royalties on sales.
|
|
Who is funding this project?
|
|
Find-a-Drug operating costs are presently being paid by Treweren Consultants.
This has no bearing on ownership or exploitation of the results.
|
|
What happens to the profits?
|
|
The Find-a-Drug trustees will distribute any profits to research. As the costs of
running the project are low, this would include a substantial part of any royalties.
|
|
Is Find-a-Drug a charity?
|
|
We are seeking charitable status for Find-a-Drug although it is likely to be some
months before this is approved. Being a charity would have certain tax advantages
once there is some income!
|
|
When does Find-a-Drug expect to generate some income?
|
|
We expect to enter into some partnership agreements during 2005 which may
include an upfront payment to transfer the IPR in the results for certain
protein targets. Thereafter Find-a-Drug will be self-funding.
|
|
Will the results be published or put in the public domain?
|
|
It is normal practice to patent the activity of molecules once this has been
confirmed experientally. This serves to place some of the results in the public domain
as well as protect the intellectual property.
|
|
What about public interest projects?
|
|
The results from public interest projects such as the Bioterrorism antidote
project are freely available to approved academic and goverment laboratories.
|
|
Can academics access the results without charge?
|
|
We look sympathetically at requests from academics for a subset of the results
as well as suggestions of certain protein targets.
|