automated DNA sequencing

Thu Oct 31 07:02:00 EST 1991

 Ellen Murphy writes: 
=>   We are beginning to think that the time has come to buy an automated
=> DNA sequencer.  This would be shared by a large number of labs; nobody
=> here is doing mega-sequencing, but all those little bits and pieces
=> start adding up.  I would appreciate hearing from people who are using one
=> or another of these machines.  Any and all opinions will be helpful.
=> 	Which chemistry, and why?  ANS: Taq Cycle sequencing, more accurate
=> 	Which machine?             ANS: At present ABI 373A
=> 	What's a ballpark minimum number of bp/week that one needs
=> to be doing to justify automation?
				   ANS: We obtain 20 KB raw data/day with
					2 9-hour runs on an ABI 373A with
					an average accuracy over 450 bases
					per lane of >98%.
=> 	What about automation of template preparation and sequencing reactions?
				   ANS: Perkin Elmer Cetus 9600 Thermal Cycler
					for taq-cycle sequencing using template
					isolated using a Biomek 1000 for
					ss-templates and manual minipreps for
					for ds-templates.  Qiagen and BioRad
					semi-automated ds-templates also
					yield comparable results but these
					are not yet automated in our lab.
=> 	Will the sequencer break down if many different people use it?
=> (Since the answer to this is obviously yes :-), the real questions are:
=> will we need a dedicated technician to run it, 
				   ANS: These instruments are not centrifuges
					which everyone can use.  I STRONGLY
					recommend a DEDICATED individual be
					in place to run the instrument, just
					as one has a DEDICATED individual doing
					DNA synthesis, amino acid analysis, and
					amino acid synthesis.
=> and which company's service department is the easiest to deal with?)
				   ANS: We have had excellent repair service
					from ABI, but do have a maintenance
					agreement to the tune of $10,000/yr.
					I STRONGLY recommend a maintenance
					agreement as the laser has about a
					one year life and costs ca. $7000
					to replace.

The following is an update to the response I posted to a similar question
in an earlier posting by J. Walton:
        Since my laboratory was the initial university test site for the
ABI fluorescent sequencing instrument, maybe my 2cents worth will be of 
interest to others on the net.

        There presently are *only* 2 automated sequencing instruments
worth considering for purchase.  The ABI 373 and the LKB/Pharmacia ALF.
Both cost around $100K and both have advantages and disadvantages.
There are other instruments for automated fluorescent or radiolabeled
sequencing, including the following I am aware of:
Dupont - no longer "really" an alternative as they essentially are out
         of the fluorescence sequencing instrument business.
Hatashi, Milligen, and others - presently in beta test.

Common Advantages:
        You never have to read another autoradiograph and enter the
data yourself into the computer.
        Pouring gels and loading samples uses the same techniques you
are familar with from radio-labeled sequencing gels.

Common Disadvantages:
        The data collected is *all* the data, including background,
and thus "@#$%-in ==> @#$%-out" (pardon my vulgarity but you get my point).
        The template *must* be pure as can be or else the above is true.
        The analysis programs leave alot to be desired because they
really do not take into account local variations from user to user.
(i.e variations in sample quality, variations in gel percentage, sequencing
reaction conditions, etc.)
        It takes some doing to optimize the DNA isolation conditions and
the DNA sequencing reaction conditions, unless you have lots of $ and
can afford to buy a DNA sequencing kit every week at $600/100 reactions.
        Proof reading and editing the output requires the purchase of
additional programs and computers.  We use the MRC programs on a
SPARCstation, not the Vendor supplied programs as they are not as good.
So think about adding another $25K to your purchase price for a SPARC2.

   Reference for the MRC programs:
   Dear, S. and Staden, R. Nucl. Acids Res.  19, 3907-3911 (1991).

        Both may be capable of dye terminator sequencing but the claims
may be over stated by both companies.
        Both are capable of sequencing off double stranded template but
again the purity of the template may be a problem.

Specific Advantages of the ABI 373:
        Well built, reliable, rapid and experienced service.
        Been around a while and the programs have improved slightly over
the years.
        Programs that are used can handle 24 samples and give data out
to 450 nucleotides from the m13 priming site fairly reproducibly, depending
on the quality of the template and reactions (we use Taq Cycle sequencing
reactions which reduces or seems to reduce the effect of contamination of
due to impure template and make our own mixed, buy our own enzyme, make
our own primers).
        ABI is addressing the issue of template purity and uniformity
of reaction pipetting and soon will introduce a robotic workstation which
for the present will pipet and perform all incubations for the DNA sequencing
reactions.  In the future they may produce a robotic workstation for automated
template DNA isolation.

        We have published to papers related to the ABI and automation of both
the isolation and sequencing reactions and a forth- coming issue of "Methods:a
companion to Methods in Enzymology" will be published with additional related

E.R. Mardis and B.A. Roe. Automated Methods for Single-Stranded DNA Isolation
and Dideoxynucleotide DNA Sequencing Reactions on a Robotic Workstation.

BioTechniques 7, 840-850 (1989).
L.A. Johnston-Dow, E. Mardis, C. Heiner and B.A. Roe.  Optimized Methods for
Fluorescent and Radio-labeled DNA Sequencing.  BioTechniques 5, 754-765 (1987).

        In addition, we are working on a series of programs which may (we
hope) do a better job of interpreting the ABI data than ABI does.  These
programs presently are almost done and will run under DOS or MAC OS and
should complement the MRC programs.

Specific Advantages of the Pharmacia ALF:
        Based on technology from Ansorge's lab at EMBL.
        Temperature control of the gel environment is excellent.
        Easy to make primers (need only one/sample since ACGT rxns are
loaded in 4 separate lanes).

Specific Disadvantages of the ABI 373:
        Poor temperature control of the gel during electrophoresis (smiles).
        Requires 4 different dyes and thus 4 different primers/sample and
it is difficult to do primer walking.  Also, each new primer (based on
sequence) requires a different mobility file for the programs to work

Specific Disadvantages of the Pharmacia ALF:
        Fewer samples/run than ABI, although the gels are shorter and
run time may be shorter the reality is one run/day and less through-put
with the ALF.
        Analysis programs are not as robust as those provided by ABI because
ABI has been producing fluorescent sequencing instruments longer and had
more time to improve their programs than LKB/Pharm. has (although their's
are improving and almost caught up with ABI).
        Very few instruments delivered in the US.  Many more in Europe.
Thus service and local experience of the tech service folks may not be
as good as ABI (which by the way really isn't that good).

        I have not addressed the issue of accuracy.  That is because accuracy
is like beauty "in the eyes of the beholder" and relates back to the purity
of the template and sequencing reactions and gel conditions.  Compression
is compression and even humans cannot figure out most compressions accurately
so why expect an instrument programmed by humans to be able to do it.

        The bottom line here regarding accuracy is that these two instruments
are at least as good as a human in "accurately" reading the gels in one pass.
Until we figure out how to reproducably obtain really pure template and how to
remove rather than interpet compressions, errors will occur.  Most humans read
gels 98-99% accurately over an extended period of time and these two
instruments also give 98-99% accuracy from the priming site to about 450
nucleotides depending upon the above factors.

	Both instruments use the standard dideoxy sequencing chemistry.  We
use Taq-cycle sequencing on the ABI 373 and end-labeled primers. ABI has a
kit for Taq-cycle sequencing but we make our own reagents and use a slightly
modified protocol with the Perkin Elmer Cetus 9600 Thermal Cycler. Although
ABI has come out with fluorescent terminators, which give reads about 100
nucleotides less than the fluorescent-labeled primers, there is a new set
of ABI fluorescent labeled terminators which will be available some time
next year that are better (at least that's what came out of 2 talks at the
Hilton Head Human Genome III meeting).  
	Most folks with the LKB/Pharm. instrument use Sequenase, but the
Taq cycle sequencing also has been used with equally good results.  

The references for Taq Cycle sequencing are:

Craxton, M. Methods: A Companion to Methods In Enzymol. 3(1) 20-26 (1991).
McBride, L.J., Koepf, S.M., Gibbs, R.A., Salser, W., Mayrand, P.E.,
 Hunkapiller, M.W. and Kronick, M.N. Clinical Chemistry  35, 2196-2201 (1989).
Protocol booklet included with the ABI Cycle Sequencing Sequencing Kit.

Bottom line:
        If you've got the money and *really* want to devote a lot of time
for a Ph.D level person to get everything up and running, then I'd recommend
purchasing the ABI 373 along with a SPARC2 and getting the MRC programs
for proof reading.  Alternatively, contact both companies and see if they
will demo the instruments for say one month in your lab.  This will give
you better first hand experience and you set the conditions for acceptance.
I would *highly* recommend this alternative rather than taking someone's
word for it.

        However, don't throw away those radio-labeling sequencing
gel plates, buffer chambers, and power supplies as you may find that
radio-labeled sequencing may be just as quick and efficient, especially
if you can read past 500 nucleotides from a priming site (with multiple
loadings or different percentage gels, etc).

Hope this was informative.
 \  Bruce A. Roe                    Professor of Chemistry and Biochemistry /
 /  Dept. of Chem. and Biochem.     INTERNET: BROE at  \
 \  University of Oklahoma          BITNET:   BROE at uokucsvx                 /
 /  620 Parrington Oval, Rm 208     AT&TNET:  405-325-4912 or 405-325-7610  \
 \  Norman, Oklahoma 73019          FAXnet:   405-325-6111                  /
 /                                  ICBMnet:  35deg 14min N, 97deg 27min W  \

 /o o\   \ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\
   o      X  X  X  X  X  X  X  X  X  X  X  X  X  X  X  X  X  X  X  X  X  X
  \O/    / \/ \/ \/ \/ \/ \/ \/ \/

        The opinions expressed are my own and as others have stated:
                "As I speek for myself and usually only to myself"
        Also, I have no commercial ties to the companies mentioned.

More information about the Methods mailing list