ABI 373A DNA Sequencer data just fades away.(longish)

Paul Morrison p_morrison at dfci.harvard.edu
Fri Feb 17 15:30:28 EST 1995

Thank you to the 14 people who responded to my sequencing problem that I
posted. Since all of them had enlightening things to say I have included
all of the responses and my original post at the end of this synopsis.

Briefly the problem: ABI373 automated sequencing, double strand "Core
Facility" templates, cycle sequence with Taq polymerase and DYE
terminators. Every 15 gels or so a dramatic loss of signal in the
chromatogram in a _group_ of samples on the gel (details below).

The reason for posting was to find out if other people had this problem
occurring in the same time frame (Oct94-present) as mine so we could blame
it on a reagent. This hasn't been ruled out but does not seem to be the
case (one response with same time frame).

Take home message from responses:
Because of the low frequency of the problem I have thought that several
factors fall below tolerances in order for it to occur. I now have three
favorites. Detergent, formamide, and X factor.
Because it is difficult to deduce even from the raw data or the gel file
whether I have a drastic loss of fluorescence or loss of resolution, both
formamide and detergent on the glass are in the running as the culprits.
For the last ten days we have deionized formamide, capped aliquots under
argon and will make new aliquots every three weeks,(if anyone has proof
this is extreme overkill please tell me). We have also increased the
rinsing procedure that follows the Alconox cleaning of the glass plates.
The problem has not reoccurred. If it does I'll be back on the net to find
the X factor.

Thanks again,

Paul Morrison
Molecular Biology Core Facilities
Dana Farber Cancer Institute
44 Binney Street
Boston, MA 02115

(this message has been cross posted to:
and the ABRF, Association of Biomolecular Resource Facilities. If you want
information about these network resources please email me.)

Allison Pinder               pinder at mail1.ciwemb.edu
W.Alton Jones Cell Science   mbcf at transit.nyser.net
walter.just                  walter.just at medizin.uni-ulm.de
Mulligan, John               mulligan at darwin.com
Thomas_C.Newman              22313TCN at msu.edu
Vahe Bedian                  DNASequence at mail.med.upenn.edu
Sheila                       Sheila at lenti.med.umn.edu
Tom Keller                   Tom_Keller at gene.biotech.wisc.edu
Mel Kronick                  kronicmn at ccmail.apldbio.com
Di James                     DI at molbiol.uct.ac.za
Steve Hardies                HARDIES at thorin.uthscsa.edu
Anthony Otsuka               ajotsuka at rs6000.cmp.ilstu.edu
Laura Livingstone            lrl at med.unc.edu
Dave Knorr                   dak at biosys.apldbio.com

_________ORIGINAL POST __________________ORIGINAL POST_____________
Subject: ABI 373A DNA Sequencer data just fades away.

Here is a problem that is starting to really bug me. It has been occurring
intermittently since Oct94 at an increasing rate. Just as soon as we think
we have figured out the problem. Whap, it's back. So a brief description
of the problem.

"Core Facility" (different sample preps) templates sequenced double
stranded, Taq polymerase, cycle sequencing, DYE labeled terminators.
On occasion we will get a gel that has _something_ wrong with it. We now
call them "fast-fade" gels. On these fast-fade gels many lanes of data
will look good but then abruptly (within 20 nucs) fade to practically
nothing. These problem lanes will usually (but not all) come in groups on
the left or right hand side of the gel. This problem has occurred in
various places. After 50 to after 400 nucs of good data. We have
contemplated just about every reason and not one of them holds up. Because
it happens infrequently and we are sequencing Core Facility DNA from
various sources this has been very hard to track down.

What the problem looks like on the:

Analysis file. Peaks are great and then abruptly drop and get the
"shakes". There is usually trailing _in front_ of the peak that happens as
the peaks spread out and quickly change to shaky waves of color.
Raw Data file. Peaks are great and then fade to nothing.
Gel file. Peaks are great and then get blurry. It looks as if the same
amount of color is there but the resolution has gone to pot.

Reasons why we have ruled out:

secondary structure:
the problem has occurred sequencing the ABI standard which shows no
secondary structure problem. Also templates will be rerun the next day,
same conditions and the problem disappears.

Template prep, primer prep:
templates from different sources fail and as above work fine the next day.

The Gel:
We have replaced all reagents many times. No correlation. 2 gels made at
the same time by the same person loaded on different 373's. One
fast-fades, one is fine. (This happened last night after we thought we got

The sharkstooth comb:
We prep this the same way. No loose acrylamide, no problems.

Sample buffer, the formamide, the blue juice:
Replaced, brand new, deionized. Still happens.

The centrisep columns:
Same lot, same technique used, one good gel, one bad gel.

The Taq enzyme, the terminator dyes:
The same lot, the same tube, one good gel, one bad gel.

The 373:
We have 2 373's. Old373 got a stretch upgrade in Apr94. New373 was
delivered as stretch Sep94. The problem has occurred on both. While it is
true that the problem has only occurred on a stretch machine, we ran
stretch gels from April to October with no problems. The problem has
occurred only once on the new 373 and has occurred 8-10 times on the old
but I feel this could easily be _not_ significant because the old one is
used more frequently.

Data collection:
The drop off does not occur at exactly the same spot (scan line) in the
gel file. It will happen in _around_ the same spot but will have enough
variation that rules out data collection problems.

Final thoughts:
Because it doesn't happen often it has been very difficult trying to pin
this down. It may have multiple reasons and they all have to be _below_
tolerances in order for it to show up. If this is the case then it may be
a reagent or column spec. that has changed since October 94.

Has any other lab seen this problem? Anything close to this problem?

Thanks for any help or suggestions and if you know another owner of an ABI
373, please give them a copy of this.  -Paul

Paul Morrison   Dana1030
Molecular Biology Core Facilities
Dana Farber Cancer Institute
44 Binney Street
Boston, MA 02115

p_morrison at dfci.harvard.edu

______END OF ORIGINAL POST____________________________________

responses in order of of reception:
Tom Keller   Tom_Keller at gene.biotech.wisc.edu

RE>ABI 373A DNA Sequencer data just...          2/6/95
We have seen various, intermittant, problem as well. In our case,
however,though we didn't want to believe it, it ended up being one or more
of the postcycle sequencing steps. I.e., getting the fluorescent extension
products to the well, or a combination of these steps. The most frequent
cause is the formamide. Buying deionized formamide is not good enough.
We've done the experiment. Freshly deionized formamide works, month old
deionized formamide,stored at 4#161# gives short reads. We have also had a
case where not paying attention to getting the entire dried sample
dissolved in the loading mixture gave intermittant problems. Finally, we
have seen samples were poor recovery from the method used to remove
dye-terminators gave poor results. It is unlikely that an instrument
problem will fix itself. There's the rub.

Sheila  Sheila at lenti.med.umn.edu

i havent seen your intermittant problem with signals suddenly fading out.
that the problem occurs with the abi standard, and that it can be both
presentand absent on the same gel depending on the lane, makes me think
what you're seeing is the result of something going on with the centricep
columns. do you have two lots pooled together? or possibly people spinning
them at different speeds.by the way, we've found that fine sephadex g50 (1
gram/16 mls ddwater) works well in recycled centricep plastic columns. we
make up a new slurry every 3 weeks. it beats paying $2 plus dollars per

Vahe Bedian     DNASequence at mail.med.upenn.edu

I can't say that I have encountered the exact same problem, but let me
address some possibilities. We have two relatively new 373s with stretch,
and the one difficult to track problem ended up resulting from the angle
at which plates were held while pouring the gel. If too verticle,
hydrostatic pressure caused some separation of the plates near the bottom.
This resulted in lower resolution (overlapping peaks) because the focal
point of the laser was not exactly in the middle of the gel. We generally
had a good start on chromatograms, with a generally fast but not sudden
drop off of signal, and waves of lower-higher resolution: ie ~60 nucl
would look OK,then lower intensity and broad peaks would appear, then it
would get better again. Another manifestation that correlates is
variablitly in electrical characteristecs and spacing. Now we clip the
plates on each side near the bottom, which does not stop the spacers from
getting thoroughly wetted, andget more consistent results. I don't know if
your problem is related to this, but since it is an affliction of the
whole gel, I thought I'd mention it.
Another thought that comes to mind is a malfunction of your cycler. If it
stops cycling correctly after a certain point, you would expect exactly
the results you describe.
Good luck,

Thomas_C.Newman 22313TCN at msu.edu

Dear Paul,

  Does your problem show up on the gel image as a green smeared
background.We have been having similar problems, and in one case have
found the problem to migrate with the gel plates.  ABI recently changed
vendors for the glassplates  and my gut feeling is that the quality is not
to the old standard.  If you would like to discuss this over the phone,
you can reach me at517-353-0854. We will have both the ABI field service
engineer and the field applications specialist here Tuesday (Feb 7).  So
if you could call me in the AM (after nine EST) I would appreciate it or
send me your phone # and I willgive you a call when I get in.  Maybe we
can join forces and get some of our problems addressed, if not solved. 
The applications specialist is coming to "show" us how to pour a gel!  We
have only poured 3 gels a day, 5 days a week for a year and a half...I
guess we are slow learners.

Tom Newman
DNA Sequencing Facility
Michigan State University
John Mulligan   mulligan at darwin.com

We have seen something that looks something like what you have described. 
My current theory on it was that it was due to residual detergent on the
glass plates (see Biotechniques Vol. 15, No. 5, p. 840).  You might try
other detergents, more rinsing...
John Mulligan
Darwin Molecular Corp
mulligan at darwin.com
Walter Just     walter.just at medizin.uni-ulm.de

Dear Paul Morrison

you encountered severe problems with the sequencer. When I was reading
your message, I noticed that you use "blue juice".I think this may be a
problem since you may not use any dyes in theABI373a.We also had and still
have problems with the ABI. The analysis ofthe data often results in high
level peaks at the beginning of thesequence and then immediately decreases
to +/- zero. When we performa manual "CALL BASES", we obtain all of our
data. Mystery?!
Best wishes
Walter Just

Allison Pinder  pinder at mail1.ciwemb.edu
Paul- I am in the DNA Sequencing core facility at the Carnegie Institution
Dep't of Embryology in Baltimore.  I just recieved a copy of your ABRF
memo re fast-fade gels courtesy of Betsy Nanthakumar of Johns Hopkins.  I
have seen this phenomenon crop up intermittantly since summer of 1994. 
It's starting toreally bug me,too.I have only one 373 in the lab.  This
problem has occurred on the instrumentin its original configuration with
24cm WTR and, since we upgraded to stretch,on 34cm WTR gels.   The same
thing can occur with a wide variety of samples:Different templates,
different vectors, different thermocyclers, prepared by different
people-all subject to the same effect.    Like you, I have replaced
reagents and changed suppliers several times over. The EPT doesn't
indicate any electrophoresis problems, and two ABI servicemenhave told me
the machine is fine and that I have a gel chemistry problem.  I have
changed glass plates 4 times in the past 8 months and had the same problem
at least once with every pair.   The problem has occurred with two
different software versions.  Sometimes the problem will disappear after
oneof these changes, but turn up again later.  I have been at my wits' end
over this.  The problem had been gone for about 2 months, then showed up
again in some samples on each of two gels last week.   I am in agreement
with your thought that there may be multiple causes, all of which must be
below tolerance to produce the effect.  My strategy now is turning toward
getting ABI to take notice and expend some real resources on solving the
problem.  Our best chance of getting them to do this is if several people
are complaining-loudly- about the same thing.  Would you let me know if
you hear from others having the same problem (I'm not on the ABRF network
now)? Hope we can communicate on solving this in the future. Thanks-
                                          Carnegie Institution of Washington
                                          Department of Embryology
                                          115 W. University Parkway
                                            Baltimore, MD 21210
                                             (410) 554-1207

Anthony Otsuka  ajotsuka at rs6000.cmp.ilstu.edu
Although we have not used fluorescence-based machines, have you considered
a problem with the amount or quality of the chain terminators?  It sounds
as though the gels are working properly since you are obtaining discrete
bands.  It seems  as though your reactions are terminating prematurely. 
This could be due to an incorrect ratio of dideoxys to deoxys or to errors
in pipetting or degraded nucleoside triphosphates, etc.  Also consider
anything that could be killing the polymerase, e.g. temperature, organic
solvents, lack of magnesium, etc.  I would suggest calling the company and
obtaining a good stock of control DNA template.  Most of our sequencing
problems result from bad template.  At least with a uniform source of DNA
you can rule out that variable.
          Good luck, Tony Otsuka
Mel Kronick     kronicmn at ccmail.apldbio.com
We saw your note on the net this morning regarding gels.  I can tell you
about something we have seen here that is similar, but not identical.  We
have seen signals fade away and come back. As you observed, it was
difficult to correlate with other parameters.  One thing we found helped
significantly was to pre-run the gels before loading (for at least 15
minutes, preferably more like 30 minutes).  We realize that many people
get away without pre-running but, as you noted, the problem probably only
occurs when several sub-optimal conditions in the gel occur at the same
time.  I would be curious if my comments have any  bearing on your
problem.  In any event, I'll ask around some more here before  we go back
East. We can discuss more when I see you next Tuesday.  Hope all  else is
going well.

Mel Kronick 

Di James        DI at molbiol.uct.ac.za

Dear Paul,

This might help. We do not have a auto sequencer, but doing manual seq. I
have noticed on occasion the same phenomena. I pinned it down to the
acrylamide. The students especially, leave traces of detergent on areas of
the plates and the matrix becomes weak or polymerisation is affected in
that area and the bands 'just fade away'. 
Di James

(Sen Tech Officer)
Micobiology Sequencign Lab.

Steve Hardies   HARDIES at thorin.uthscsa.edu

Paul Morrison writes

[extensive description of a sporadic abberation affecting
sequence in some regions of an automated seq. gel]

My experience is limited to manual seq., but I am struck by the similarity
of your problem to the glycerol artifact.  That is, the sporadic nature
(related to the way people sporadically decide more enzyme is better), and
the position and nature of the artifact (bands spread out but not missing,
and only high on the gel) sounds just like the glycerol effect.  I don't
know if your sample work up removes glycerol, but if you're not using
glycerol tolerant buffer, you might try it.   
More generally, it might be anything in the samples that varies from
sample to sample, runs as a high broad band on the gel, and interacts with
the DNA running along with it.  You say that the same template will give
the problem one day and not the next.  I would suggest taking such a
template and doing a series where the amount of template used varies from
way too much to less than usually used.  If the problem reproducibly sets
in on the high end of the  template, then you're bringing something in
with the template that's causing the problem, and the sporadicity is
related to how much this substance gets concentrated on the way to the
gel.  A similar strategy aimed at other components (primers, enz) might
finger the  problem.   
Good luck; sounds like a tough problem.
Steve Hardies, Assoc. Prof. of Biochem., Univ. of Texas HSC at San Antonio
Hardies at uthscsa.edu

Laura Livingston    lrl at med.unc.edu (Laura Livingstone)

Hello from another Core Facility:

        Have you made any progress with your problem?  When you have
fast-fade, is it a resolution problem where the peaks broaden to low
signal or is the signal of the larger fragments gone or reduced?      I
don't think we have had exactly the same thing but we have had major
problems with resolution quality of peaks and, like you, can "prove" that
all imaginable sources are not causing it so have been unable to track
down the source of the problem.  We had alot of problems after Stretch
upgrades but both of our machines were upgraded within a month of each
other so maybe it was the upgrades just coincided with whatever the
problem is.

Laura R. Livingstone, PhD
UNC-CH Automated DNA Sequencing Facility
U. North Carolina
Dave Knorr    dak at biosys.apldbio.com (Dave Knorr)


I read your post last week about the problem(s) you are having with your
373 runs.  I'm not currently sequencing much, but your observations
sounded vaguely familiar, so I contacted a bunch of folks here who are
developing some of our sequencing chemistries. 
As you pointed out there may be more than one problem going on, but the
responses I received boiled down to a couple of areas to look at.  
1)  Even though you replaced the formamide, it is importnat to make sure
it is truely deionized.  Our QC guy runs it through resin 3X, then freezes
it in 500 microliter aliquots.  He keeps those only for a couple of
2)  Yo may be having gel problems.  Recently we have noticed
inconsistencies reagent mixes from a couple of suppliers.  Bis-acrylamide
mixtures may not have consistent proportions of the two, and pre-weighed
pre-mixes may not have accurate weights.  This will throw off the amount
of crosslinking and percentage of the gel.  For in-house QC we now use
only hand-mixed gel ingredients.  
3)  Indeed the glycerol present in some of the reactions will cause
problems with smearing at higher molecular weights similar to what you
described.  We have occasionally seen situations where gels poured
sequentially by the same person may not be consistent.  Even more rare is
the situation where only a portion of the gel is good. 
I hope this helps.  You may want to call Technical Support at ABD about this.

Dave Knorr
Perkin Elmer/Applied Biosystems
Agricultural Applications
dak at apldbio.com

__________END OF RESPONSES___________________________

More information about the Biochrom mailing list