An interview with Keith Davies of Find-A-Drug.
In response to many questions I have fielded from our members, I decided just to ask the guy in charge some questions to help everyone understand what the Find-A-Drug project is all about. Who better to ask than the guy in charge, right ?
The following is a transcript of an email interview I did with Mr. Keith Davies regarding some of those questions.
: Could you describe in laymans terms what the point of the FAD project is ?
: Most infectious diseases are caused by bacteria or viruses and many diseases
are associated with human proteins not performing their proper role. We are
trying to find molecules which inhibit the role of certain targetted
proteins. Such molecules are known as hits and have the potential to be new
drugs. For instance, a drug molecule might inhibit a virus from replicating
or cancer cell from growing.
: How did you get involved in the FAD project ?
: The first Internet computing project THINK was used for was the Oxford
University/United Devices project. United Devices were looking for a public
interest project which might promote their distributing processing software
and approached Oxford University. At the time I was an Honorary Fellow and
suggested a large scale Virtual Screening project and choose the original 12
anti-cancer protein targets. Later it became apparent that United Devices
were not interested in utilising more recent (and better/faster) versions of
THINK and I set up Find-a-Drug. This also provides a framework where we can
more easily monitor progress as well as collaborate with more research
: Can you describe to us what you do for the FAD project on a day to day basis ?
: Like most distributing computing projects, Find-a-Drug is under-resourced.
Fortunately the project runs without day to day intervention unless there is
a major server outage. Selecting a suitable target protein and creating a
corresponding query can take days. Input from collaborators is useful but
finding suitable experts and meeting them also takes time! Unfortunately,
working with partners is often delayed by more mundane but urgent tasks such
as answering e-mail queries or posting on the forum. I also play an active
part developing new versions of the software.
: How helpful is the FAD project in actually helping find new cures and drugs ?
: The time-scale for drug development and clinical trials is very long and the
cost extremely high ($10 million dollars is rarely adequate). We have had
some early success finding molecules which exhibit anti-cancer activity and
anti-HIV activity in the US National Institute of Health (NIH) laboratory
tests. We do not publish which molecules or how many have been found as
this might compromise agreements with potential partners to fund the develop
such initial hits into drugs. However, these results which tested only a
small subset of molecules, lead us to expect 10-30% of the hits to exhibit
the desired activity when tested in the laboratory.
: To the best of your knowledge, approximately how many servers is FAD using
currently and how many computers are running the FAD client for you now ?
: We publish the number of successful registrations (unique member numbers)
and members (unique Nickname and e-mail address) - ignoring those which
never return any results. However, such numbers can easily be misleading
because in a local network installastion many PCs can share the same member
number! In addition, like most other distributed computing projects, just
because a member returns some jobs does not mean that they continue to run
the software 24 hours every day! Personally, I find the number of computing
days per day one of the best measures. Unfortunately, this is difficult to
use comparatively because other projects use elapse days rather than CPU
: What are the long term goals for the FAD project ?
: At a simplistic level, our goals are to find better drugs for a variety of
serious and life-threatening diseases. In practice, it will be many years
before we know that this has been achieved because of the timescales for
drug discovery and development.
: How long is the FAD project going to run ?
: There are approximately 28,000 structures in the PDB (http://www.rscb.org
and in additional high quality models of homologues (which have similar
amino-acid sequences) can be built. It is necessary to have an
understanding of the role of the protein before one can judge the value of
trying to block or inhibit the role of that protein. Nonetheless, there
are probably thousands of viable protein targets and it will be many years
at the current level of membership before we will have hits for most of
: Could you tell us the basic difference between FAD and some of the other
Distributed Computing projects?
: We try to put the science first and avoid separating the operation of the
project from the science. Thus we can and do review the results as they are
returned and can make decisions quickly based on such results.
Unfortunately, this sometimes means that other very public and visual
aspects of the project suffer. From time to time we get negative feedback
on the website and the GUI (but then doesn't every programmer?)
There are also a number of subtle differences like the fact that we favour a
low level of redundancy. (In our opinion it is pointless comparing
duplicate sets of results for all jobs because they are NOT identical due to
rounding errors. In other words, we need sufficient redundancy to perform
some sample validations). We also have a high level of features: Windows &
Linux support; Local network support; Import/Export to PCs without Internet
connections; flexible configurations etc.
: Thanks for letting Linux users help out. Is there anything that Linux
users in particular could do to help your project out, aside from the
obvious of running the FAD client ?
: Less experienced users of Linux probably need encouragement to try it.
: Who are "Treweren Consultants" and what part do they play in FAD ?
: Treweren Consultants Ltd is the company that develops the THINK software and
funds the direct costs associated with the project. The Find-a-Drug project
is run on a not for profit basis. We have made a commitment that any
profits will be distributed for further research via a charity.
: Who benefits from this project?
: If we achieve our research goals, then individuals who would otherwise die
or suffer serious illnesses will have the prospect of new and better drugs.
However, drug development and clinical trials take many years and cost tens
of millions of pounds. This means that partners have to be found with deep
pockets and an appreciation that most drugs fail to make it to the market
(the most recent figures suggest that 99% of new drug candidates fails
during development or clinical trials). Our arrangements with such partners
means that if a molecule achieves certain milestones in the development
processs, revenue is channelled via a charity to fund basic research.
: Who owns the patents in any new drugs that are found?
: It is usual to patent the experiental findings in the laboratory. This
means that a collaborating partner files a patent when and if they thought
it was appropriate to protect their discovery so that they can justify
investing more money developing a drug by preventing competitors exploiting
the discovery. (FYI: Automatic processing of data does not generate IP
which can be protected by International patent laws).
:Why do you think people participate?
Most members are motivated by the desire to find cures for some nasty
diseases. Many people give money to cancer charities but donating computer
time is something which individuals can monitor their contribution and the
progress of research. Friendly competition in teams can also make
participation more fun.
:Can any-one access the results?
: In some public interest projects such as anti-Bioterrorism, we have stated
that we will make the results available to any bona fide academic or
commercial research group. We welcome suggestions of targets from
scientists and make the hits available to them. We only make results
available to other parties once we are satisfied that they will make proper
use of them and enter into a partnership agreement which complies with our
Related links for usalug.org members:
<< Our DC forum
The Quick Guide to setting up the FAD 1.25C Linux Client
FAD installation bash script
Team USALUG's Stats
Another nice Stats page
My thanks to Keith for taking time to answer some questions. If your running Linux, and would like to join our team, and help the FAD project, feel free to join us. We welcome all new members and if you need help getting things to work with the Linux client .... just post in our forums here. http://usalug.org/phpBB2/viewforum.php?f=92