>From: bergman at zeus.phri.nyu.edu (Mark Bergman)
>>What are your thought on using GCG with no local databases? Do you
>think it's feasible to rely upon browser- or email-based search
>services for all non-blast searches?
Depends on what kind of searches you do, an how quickly you want
your results. For example, we recently looked at running TBLASTX. On our
system here at UGA (DEC Alpha 3000-700; local databases), the turnaround for
a search with a 2kb query against GenEmbl was about 4 min. Using our setup
of GCG's BLAST as a client to NCBI's BLAST server, we had to wait about 90
minutes before the search ran and the results were returned.
Note also that your use of databases isn't limited to similarity
searches. Our users frequently run Pileup, specifying sequences with
a list file containing pointers to database entries. If Lookup or
Stringsearch is used to get an initial list file, it may only take a
little editing to get ready for doing the analysis. Without local DBs,
you'll have to download and reformat each sequence you want before being
able to do the alignment.
The convenience of reading sequences from a database is there for
a variety of programs. Note that this approach also guarantees you're
working with the 'right stuff'. In contrast, if you've got files with
sequences from previous database releases, you can find yourself working
with data that isn't the same as what's currently available. Sequences not
only get corrected, some have even been removed.
> We've got Mac users with fast (T1) 'net access.
The speed of access wont matter at times when you're waiting for
the server queue to get to the search you want to run. (And what about
the times when the server is down or unreachable - it has happened.)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
/ Michael J. Weise, Ph.D. \ Biological Sciences Computing Resource \
( weise at bscr.uga.edu \ University Computing & Networking Services )
\ _ _'Tis_only_me_speak'n._ _\_ _UGA Athens GA 30602(706) 542-1409_ _ _/