gene identification troubles
rebsilver at usa.net
Tue Mar 3 08:21:04 EST 1998
R.D. Haigh wrote:
> Hi Tammy
> I'll present my background first - I'm a bacteria man so some of my
> info will be biased that way - I never did zoology either what sort
> of bug(?) is L.littorea ?
Hi Tammy,I have mostly done Drosophila and plant work so far.
> My answers
> 1.I can't help you here other than to agree it seeems odd you should
> be picking up RNA genes in this sort of screen.
Guess: could it be that more RNA genes are transcribed as part of the stress? The
organism maybe needs to make more because it can't repair what it has??
Definately be very wary of low homologies at the sequence level.
> 2. The age old problem - you're enormously overjoyed that they're novel
> but what the hell is it anbd no clues available. The obvious next step has
> got to be trying to make knockouts - if these genes are only required for
> stres responses you might expect mutants would be non-lethal in
> permissive conditions, though probably showing phenotypes which might
> help with working out function. This all requires knockout
> strategies which may not exist for L.littorea (or you may not have
> them) so the next best option is overexpress your gene product, make antibodies and
> check the expression pattern either in whole organisms or cell
> By my calculations you are seeing 25-40% (?!*) of your clones are
> coming up with homologs - how good are these? do any of them make any
> sense ?
I agree about the expression patterns.But I think it is very important to do your
computer biochemistry at this stage.
If you have something that looks like an amphipathic molecule, you might want to check
it for protein protein interactions. There are many structures that might indicate
your protein has either an enzymatic function or transcription factor (see point 5).
> 3. I haven't seen any (but can't say I've looked hard) - generally
> everything goes into Genbank and if it doesn't, it is probably because
> someone doesn't want it available so you won't find it anyway.
I doubt it as well.
> 4. Sorry not my subject
Drosophila and Plants are 'normal'.
> 5. Protein translations are usually very useful - when looking across
> species (especially in bacteria) DNA matches are usually much poorer
> than protein.
You can submit your sequence and, if you ask it nicely, it will translate it in all
3 reading frames, and compare that to its protein data base.
This is where the wobble really matters. If the protein is important, chances are the
organism doesn't care which codons are used as long as they get the AA sequence right
or similar, the protein will still function.
> I tend to favour blast searches using a translation
> either against the protein database or even a translated version of
> the entire genbank DNA database - all available on the ncbi blast pages
> Have you tried any prediction programs - you should be
> able to get a fair idea of the cellular location of your protein but
> do note they are not infallible - PSORT at psort.nibb.ac.jp is
> very good for bacteria but i can't vouch for eukaryotes. there are also
> lots of useful programs (and some crap ones - you'll have to work out
> for yourself) at expasy.hcuge.ch/www/tools.html.
Commercially:I have used MacVector. It is great, but pretty expensive.
Gene Jockey is cheaper, and it works. The graphics aren't as nice when you want to see
> I hope my ramblings are some help
More information about the Methods