In article <8DD3D27DC6A5D111823200A0C955FB4004CBA0 at drucker.niehs.nih.gov>,
Staffa.Nick <staffa at niehs.nih.gov> wrote:
>Some folks at this institute are unhappy with GCG in its SeqLab incarnation.
I like SeqLab, but see my rant below about SeqWeb
>We are hoping that SeqWeb will cheer them up a bit, but the grudge has set
Well, for $2500 you get something that only works on Netscape
Communicator or IE4 (which immediately rules out well over half of my
users, who have old Macs that couldn't hope to run Communicator - they
struggle with Netscape 3)
It has absolutely no support for running jobs in batch, so you as the
sysadmin have no control over how many jobs are running on your
machine. This means that a malicious user could just overload your
machine with two dozen fasta searches and you couldn't stop it.
By contrast the GCG command line and SeqWeb interfaces (and Martin
Senger's W2H) can be easily integrated with batch/at, or with a little
more effort with more sophisticated load balancing software -- we use
Platform Computing's LSF. Consequently on my server there is never
more than one job running per CPU, which means the machine almost
never swaps, and the overall throughput is much higher than it was
when we used not to batch things properly. I refuse to go back to
those bad old ways, so SeqWeb is totally out. I will stick with W2H.
>We've had GCG so long that some scientists even grumble that they
>weren't consulted in its choice. So now my job, as lowly support
>person, is to present them with all the options and alternatives and
>let them choose.
Well, we do that to a certain extent on this machine. For sequence
assembly they can use Staden, for searching we have fast parallel
implementations of both BLAST and FASTA, for database lookup there's
SRS 5, and for multiple alignment we have clustalw. That's covered
about 95% of what people spend most of their time doing with GCG here,
and despite the performance improvements, especially with BLAST and
FASTA, they still choose the GCG versions frequently, because of the
consistency of its user interface.
>What DNA and Protein Analysis software are you aware of that would do
>some of the things that GCG would do and perhaps be friendlier?
clustalw is friendlier than pileup, and a lot more powerful. I think
of especial value is its ability to use structural information to
dynamically change the gap penalties along the length of a sequence;
obviously a gap inserted in an alpha helix is going to wreck it, for
example, and this is something pileup does not consider.
Staden is a superb DNA sequence assembly package (as someone else said
in this thread), but it does require X Windows. With the prevalence
of Linux and so on around these days, though, this should not be a
I think when EMBOSS appears, GCG may find they have shot themselves in
the foot too many times, and large numbers of people will jump ship.
Dr T J R Cutts Tel: +44 1223 333596
Dept. of Biochemistry, 80 Tennis Court Rd.
Cambridge, CB2 1GA, UK