Dear GCGers:
Unfortunately due to budgetary constraints I will be unable to make
it for the GCG users' focus meeting. I hope someone reading this can
bring up the following points at the discussion session:
1) I have never liked WPI, chiefly because of the klunky X-terminal
emulators. Since not all users have a Sun workstation at their disposal,
how about native-Mac and native-PC configurations? Alternatively, html
forms pages usable with Netscape Navigator.
I'll keep an open mind about the GUI; I have seen a little of GDE and
it looks like a vast improvement. But I won't be using WPI in my courses
until it becomes SeqLab.
2) The BLAST shell works well with the NCBI server. Other similar shells
for other remote servers would be useful, in particular one for the BLITZ
server at EBI.
3) The "gene search by signal" really needs to be beefed up. The book
"Sequence Analysis Primer" has all the matrices but there is no GCG
implementation. We are all aware of the limitations of these methods, but
it would be useful to scan sequences for potential promoters, ribosome
binding sites, etc. Other gene finding programs based on codon usage and
embedded motifs, especially in sequences with frameshift errors, would be
helpful. FRAMESEARCH is a good start.
4) Why must FASTA continue to impose a particular matrix (and a different
one in Version 9.0) on the user? A menu with a proposed default would be
much more flexible and would impose only one more <CR> on the naive user.
Also, TFASTA needs a matrix distinct from FASTA; one that penalizes
alignments to closed reading frames (by penalizing alignments of any amino
acid to a stop) and reduces the (too) high scores of, e.g. W-W identities
which often turn out to be just TGG-TGG at the DNA level and gives things
like "90% identity in 10 aa alignment". BTW, I have been using a homebrew
matrix for TFASTA for years with excellent results - significant
alignments sometimes come up from 120th to 20th place while the useless
alignments to closed reading frames are avoided.
5) The protein secondary structure prediction section is showing its age
and shells to modern methods, i.e. ProteinPredict, would be useful.
6) With the increasing genome data, shells to permit iterative e.g.
FASTA's on a list of sequences would be useful (although some system
managers may object!).
7) Can the EGCG programs be fully integrated into GCG like FOLDRNA and
FASTA have been? Many EGCG's simply add extra command line switches.
Others, like TPROFILESEARCH, have been very useful but getting them
installed and reconfiguring their parameters is not always easy.
8) Better interfaces to other databases like TFD and EPD would be
preferable to things like %Findpatterns -patterns=data:tfsites.dat where
the database size always seems to be outrunning the program parameters.
9) In summary, more emphasis on implementing new sequence analysis
algorithms rather than on the GUI is necessary.
Excuse the large bandwith and please don't interpret any of this as a
flame. I have been very happy with GCG and would like to help it remain
the comprehensive tool that it is.
Paul
******************************************************************************
Paul H. Roy Phone: +1 418 654 2705
Departement de biochimie,FSG FAX: +1 418 654 2715
Universite Laval E-mail: proy at rsvs.ulaval.ca
Quebec, QC G1K 7P4
CANADA
******************************************************************************
