About Compressions with Sequencing

pmiguel at bilbo.bio.purdue.edu pmiguel at bilbo.bio.purdue.edu
Sat Jun 25 12:35:14 EST 1994

In article <carterd.772491166 at news.ohsu.edu>, carterd at news.ohsu.edu (Darrick Carter) writes:
>Does anyone have a good suggestion about how to sequence a very G-C rich
>genome? We've tried formamide in the gel and deaza-guanine, but even
>these don't seem to get rid of the compressions. Any help?
>	Thanks already,
>		Darrick

  This depends on what you mean by "compressions".  Are these artifacts 
caused by secondary structure of the template, of the product, or both?
It seems that even very experienced sequencer's I've met have been a 
little vague on this point.  
  If I see "strong stops" or "fall-offs", that is, bands across all four
lanes, then I generally presume this is caused by secondary structure of
the template.  This structure can cause the polymerase to "prematurely
terminate" which means it stops extending the strand even though it has not
incorporated a dideoxy nucleotide.  If this is my problem I can add as much
deaza-G to my reactions and as much formamide to my gel as I like and it is
unlikely  to have much effect.  Secondary structure in the template can be
dealt in two ways:  1) eliminate or diminish the secondary structure so
premature termination won't occur or 2) modify premature termination
products after terminations so they won't be seen on the gel. By using
elevated termination temperatures (I always run my sequenase terminations
at least at 42 oC -- but Sequenase is touted to be able to withstand 50 oC,
or even 60 oC in the presence of large amounts of glycerol) or single
stranded binding protein secondary structure of the template can be
eliminated or diminished.  If even 60 oC is not doing the job it is 
possible to use thermostable polymerases (though I guess they must be exo- 
to work for sequencing) in combination with Sequenase or instead of 
sequenase.  A (probably more expensive) alternative is to used terminal 
transferase to add a large number of nucleotides to the 3' ends of any 
strands with 3' hydroxyls.  "Real" sequencing products will end with a 3' 
deoxy and not be extended but premature terminations will be extended by a 
few hundred nucleotides and will run much higher on the gel, probably as an 
undetectable background smear.
  I should add that another source of bands across all lanes is "dirty 
templates".  I'm not sure what this mean exactly but it is generally used 
to refer to miniprep DNA which hasn't been subject to secondary 
purification by phenol, ammonium acetate cuts and/or silica 
binding/washing/eluting.  One complication here is that in my hands (though 
not in everyones) many host strains of coli do not produce good quality DNA 
for sequencing -- no matter how much purification is performed.  HB101 is a 
good example of this.  This is why I always use an endA- strain.  (Actually 
I always use DH5a, but I believe that it is the endA- characteristic of 
DH5a that causes it to produce miniprep DNA of high quality for 
  All that I've written about above I don't consider to be "compression".  
Yet since premature termination and/or dirty templates are often associated 
with or confused with what I consider real compression I thought it 
important to discuss.
  I define compression as a sequencing gel artifact caused by product
secondary structure.  I recognize it by bands running very close together,
possibly out of order AND followed by an "expansion", that is, bands spaced
quite far apart, generally in proportion to the length and severity of the
compression.  Often a minor compression will be resolved perfectly on the
other strand.  Even major compressions will be compressed most severely at
a different segment of the compression on each strand so that careful
reading of the gels for both strands will give a fairly accurate reading of
the compression -- or at least of everything but the middle. 
  Though long, severe compression can be very vexing to resolve, it is the 
short compression, which causes two identical bases to run as one, that is the 
most frightening.  This is because they can lurk anywhere in my sequence, 
unnoticed by me yet causing a frame-shift that can totally skew my analysis 
of the sequence.  Vigilant second-strand sequencing is necessary to find 
  To resolve compression one may: 1) destabilize the product secondary 
structure by running hot gels or introducing formamide into the sequencing 
gel or 2) introduce nucleotides into the sequencing polymerization rxn which 
will produce much weaker, or no secondary structure when incorporated into 
the product.  
  I get the impression that sequencing gels were once run much hotter than 
they tend to be these days.  Last time I read the red book it suggested 
that one knew if one was running a sequencing gel too hot if the plates 
cracked.  So I think compressions are more problematic now than they used to be 
because fewer are stable at 65 oC than at 50 oC.  But heat, even if not 
enough to crack plates, can warp them and also tends to cause bands to be
more blurry.  
  Formamide, I`ve found, works quite well.  But it may not, by itself 
completely resolve a compression.  I tried once to deionize cheap formamide 
with a mixed bed resin.  The gel I poured would not polymerize -- so I 
always pay Clonetech's exorbitant price for deionzed formamide.  Also, I've 
only ever run an 8% formamide gel so I needed my compressions close to the 
primer site.  But this is a good idea in any case -- the closer one is to 
the primer the better the chance for resolving the compression just 
because the band spacing will be greater.
  dITP is said to have no capability to produce secondary structure when it
replaces dGTP.  Unfortunately it tends to cause premature terminations
which make it unsuitable for a simple replacement of dGTP.  Instead one
will generally run 2 sets of rxns side by side on a gel (same template)
using dGTP in one set and dITP in the other.  dITP is also especially
sensitive to Sequenase's tendencing to run the polymerization reaction in
reverse at times -- so it is important to include IPPase in sequencing
reactions which utilize dITP.  Fortunately IPPase is so cheap in
comparision to Sequenase there is little reason not to include it in all
reactions.  Finally, by chasing dITP reactions with terminal transferase to
get rid of premature terminations -- it is said to be possible to use just
the dITP reactions and dispense with the dGTP set. 
  deazaGTP can produce some secondary structure but it is reputed to be 
much less than dGTP.  But it doesn't cause premature terminations like 
dITP, so there is no need to run 2 sets of reactions.  Pharmacia sells 
deazaATP (or something like that) which, they claim, in combination with 
deazaG will totally resolve compressions.  I've never tried either.
  Finally, a brief note on cases where there premature termination and 
compression.  Just realize that they are seperate problems and must be 
dealt with in different ways.  Clear up the premature termination problems 
first, if you can -- because they will definitely make your compressions 
much more difficult to resolve.  


More information about the Methods mailing list