In article <1992Jun17.182221.16335 at sifon.cc.mcgill.ca>, francis at monod.biol.mcgill.ca (Francis Ouellette) writes:
|> Thomas L Sims (tsims at magnus.acs.ohio-state.edu) writes:
|> >1. How do the stand-alone programs compare price-wise to the on-line programs
|> >(i.e. UWGCG or Intelligenetics IG-Suite)? I'm _not_ talking about having a site
|> >license on your local VAX or Unix machine, with a system manager etc., but in
|> >terms of (poorer) sites where one would have to telnet to a remote site for the
|> >service. Does the cost for cpu minutes for searches/sequence comparisons
|> >eventually equal/surpass the stand-alone price over the course of 3-5 years?
|>|> Well I am sure this will vary from site to site. I know that a lot
|> of institutions (this one among them) will provide "free" Internet
|> access to its reserchers, so I guess it would take a few years to
|> catch up to the price of a PC! But let me rapidly add (so to avoid
|> missunderstandings) Internet is not free, it cost, but some school
|> pay up and do not transfer the cost to its users. This may not
|> last forever.
True. But a single system costs a fortune. You surely can buy a reasonable
PC for $5000. But, you don't buy the equence database updates, and the
expertise how to use a system with it. And, you are sort of alone if you
need to reproduce results if you use standalone software. It sure is not
necessarily lethal nor desirable to go with the main stream, but do
you really spend time in evaluating 5 software packages in the PD, just
ending up in getting all of them because none does it all?
|>|> >2. How do the stand-alone programs compare in terms of difficult computational
|> >requirements to the on-line programs?
|>|> I am not sure about the limit of CLUSTALL (in PC/Gene or the PD
|> version) is, but I am sure it is beyond 24. I am sure somebody
|> will correct me if I am wrong :-)
|>
You can do very fast searches with precompiled databases, as some of
the commercial PC packages do. You can swap a VAX with unlimited CPU
if you make a 1000 bp alignment and search the database with a DNA profile.
Typically, the searches should be done on at least workstations.
|> >3. Do 386/486 machines give you a substantial speed increase using the
|> >stand-alone programs versus slower machines? I've always had the impression
|> >that the PC-programs were written for AT machines (based on economic realities)
|> >and couldn't really take advantage of the greater processing speed/memory
|> >available in newer machines?
|>|> It is very true that very few packages take advantage of the multi-
|> processing capacity of the 386 and 486, but with these higher-end
|> models you also get "faster" (the MHtz number) performance with
|> "cheap" 50 MHtz/486 on the market ... I assure you, you will
|> see a difference.
|>
Once you are talking about multiprocessor 486 I think you should ask your
nearest workstation sales rep and ask for the pricing. A mainframe package
on a pizza box is presumably easier to maintain than a 8088 ported application
on a 486 twin system.
|>|> But without wanting to get into the ball going on this subject
|> (again!) maybe a low end workstation is an alternative to a PC
|> buyer. There is a lot of "front ends" for public domain (PD)
|> software out there, GDE for example. A wonderfull piece of
|> work which lets you use all sorts of PD software.
|>
If you can make sure that
(1) you can pay for the maintenance
(2) have some net expert on hand to get the workstation set up
(3) get ONE REASONABLE (as opposed to all PD on any server) software
that would be great. I think that a low-end Indigo, Sun or DECstation with
some GCG and a genome package from the PD would get you a good start
for about $25K. Surely beyond the $5K cited abouve but a real thing.
Now shop around and ask how long you can get a cpu usage with online
services for these $25K. Take your option.
|>|> >5. OK, one more.
|>|> you are welcome,
|>
Maybe some others in the same situation would like to speak up?
Regards
Reinhard
--
+----------------------------------+-------------------------------------+
| Dr. Reinhard Doelz | RFC doelz at urz.unibas.ch |
| Biocomputing | DECNET 20579::48130::doelz |
|Biozentrum der Universitaet | X25 022846211142036::doelz |
| Klingelbergstrasse 70 | FAX x41 61 261- 6760 or 267- 2078
| CH 4056 Basel | TEL x41 61 267- 2076 or 2247 |
+------------- bioftp.unibas.ch is the SWISS EMBnet node ----------------+
-----------------------------------------
