SUMMARY -Logistics of running an ABI 373A sequencing facility

David Johnstondaj daj at uk.ac.ic.nhm
Tue Sep 7 05:14:05 EST 1993


Greetings netters,
A few days ago, I posted a request to the "Methods and Reagents" and 
"Human Genome Project" newsgroups posing various questions concerning 
the logistics and day to day running of an ABI 373A sequencing facility. 
I was inundated with replies so a big thank you to those listed below (and to 
Marion Piper from the University of Sussex who even phoned up in response 
to my posting) for sharing their experiences and offering advice. I have
comliled their replies for general info.
 
Martin Hughes 
mjgh at mbfs.biology.cambridge.ac.uk

Raj Shankarappa
bsh at med.pitt.edu

Rick Meinersmann
RMeinersmann at asrr.arsusda.gov

Jaime Prilusky
lsprilus at weizmann.weizmann.ac.il

Thomas C.Newman
22313TCN at msu.edu      

Greg Lennon 
greg at mendel.llnl.gov 

Roger C Wiegand
rcwieg at ccmail.monsanto.com

Aaron Liston 
listona at BCC.ORST.EDU    

Mick Jones
mjones at rpms.ac.uk 

Bruce A. Roe                 
BROE at aardvark.ucs.uoknor.edu

Bob DeLisio
delisior at rnisd0.DNET.roche.com 

John H McDonald
mcdonald at ravel.udel.edu 

****************************************************************************
(1) Is multi-user operation practical? If so by how many users? Or is 
dedicated technical support essential?
****************************************************************************

Yes, I would say multiuser is no big problem.  We had about 6 folks using 
it (I know that isn't many, but I see not problem other than the usual 
enforcement of clear-up regulations etc).I would, however, strongly advise 
that any "big time" users got and kept their own plates.  They can get 
damaged quite easily, and personally I would not trust anyone else to keep
plates clean.  When you consider the price of the beast, a few extra sets 
of plates are not going to break the bank.  There is nothing more annoying 
than casting your gel, preparing your samples and being ready to go, and 
then seeing a hugepeak on your plate check, due to gunk on the inside left
their by a previous user of the plates. I would not have thought you needed 
a technician for the machine, unless you really forsee a problem with clean-
up/maintainance (eg. with the Mac system, the printer we got worked on some
sort of transfer technique.  Each roll would do about 200 sheets, and then 
HAVE to be replaced.  Someone needs to keep an eye on this sort of thing, 
but hardly a need for a dedicated assistant. Again - a clean up sort of 
problem, you need to make sure the computer disc is kept quite empty - copy 
your own files once you have finished, and if you leave them on the 
computer, do not complain if they get wiped.

Yes and Yes.   It is quite practical to make it multi-user operation.
or at the least the cost of the whole setup requires it to be multi-user
to be cost effective.  In my opinion, it is very essential to have one
dedicated technical support person.  This person can be responsible for 
maintainace of the instrument, setting up the run conditions, and if
possible running the samples.  I would run it like a core facility in 
that the person who wants the sequences done will do all the reactions,
and give the responsible person the final material ready to be loaded on 
the gel.  If possible, I would also make each lab using the instrument
purchase one or two sets of glass plates, combs and pour the gel before
coming to start the instrument.  Of course this makes the necessity for
one full time person less stringent.  

One thing of critical importance with multiple users on this machine is that each user
should have his own set of glass plates, combs, and spacers.  They are 
relatively expensive, but will solve a lot of headaches.

I would suggest you to have a  dedicated technical support. We currently have
2 full time technicians working on the sequencers. The simple prepation of
the gel takes time. We do have an ABI robot for the sample preparation.
Even so, the dedicated people always have something to do. On the other
side, I see the equipment too delicated to be passed from hand to hand.
The facility is multiuser in some sense. We are receiving data from
different departments of the Institute.

I have observed that those labs which have a dedicated technician to run 
and care for the machine end up using the machine and getting good 
results.  Other labs with an open user policy usually have an ornamental 
machine.  Try to get a dedicated technician to do the runs. We find the 
greatest problem is that when other labs bring in samples the DNA is not 
properly quantitated.  For these machines too much DNA is just as bad as 
too little.  We require that all DNA is quantitated by OD260/OD280.
Additionally we highly recommend the use of either promega or quiagen 
column purification for the DNA.  The results are much more reliable. We do 
sequencing campus wide and we organize that each lab has a folder 
for submission of samples which they fill out, specifing, template name, 
primer, type of sequence chemistry, etc.  We then put them in our 
sequencing queue and run them with similar samples.  We try to have a 1 
week turnaround time.

In my opinion (& I am only a user, not the operator) a dedicated 
technician is essential.  However, I do think that students and researchers 
should have the opportunity to do "internships" learning from the 
technician how to operate the machine.  Our technician dos not want others 
to set up the reactions, but I could see a place for that in student 
training.  

My own opinion is that it is better to have one or two technical people run 
the machine, i.e. make the gels,load the samples, etc.  The data can then 
be sent to another computer. People requiring sequences can always perform 
the reactions themselves,  ppt the DNA ready for loading, or have the
dedicated technician do it all.

I don't think at 80,000 pounds you can afford to have too many people using 
the instrument.  I find with multi-user equipment you always find it in a 
mess when you want to use it, and nobody owns up to being the last user who 
made the mess.

A multi-user operation is NOT practical unless you or someone has extremely 
tight control over them, i.e. in my lab we have 4 ABI's and each of my 
students take turns (after a month or 2 of training)running the 373A's.  If 
you have a single ABI 373A, then I STRONGLYrecommend dedicated technical 
support!!  Part of the reasoning behindthis recommendation is that cleaning 
and pouring plates takes lots of practice, actually loading the gel is no 
problem BUT making sure the plates are clean is.  Also, it seems to take 
some skill in handling these plates (remember they are $1000 US/set) so 
break one and it really costs.

****************************************************************************
(2) Prefered specification of networked Macs, VDU size, RAM, Disk size etc
****************************************************************************
 
It takes about 20 mb for each gel file and analyzed data.  We store all
our gel file and analyzed data on optical disks.  These disks go for
about $75 each.  It may be beneficial to have each person using the 
sequencer obtain an optical disk drive and disks.  You cannot afford to
store any of the data in the hard disks.  One additional thing you need
to consider is that it does take about 30 minutes for data transfer 
from machine to an optical disk and more if it is networked (depends)

If your machine will have a lot of use then I would recommend obtaining a 
second Macintosh computer, the Centris 650 is probably currently the best buy.
The computer that comes with the sequencer can be connected to the new 
one with just an Appletalk cable and you can put the printer that came 
with the sequencer on the new computer ( let the computer on the sequencer 
just run it and collect data, use the new computer to analyse and base-
call).

While you've got the money, buy a Sun computer with 2 gig hard disk
and get the MRC/WashU programs XBAP and TED.  Alternatively, Sequencher
and DNAStar might provide you what you need and are Mac-based.

****************************************************************************
(3) Raw data backup policies.
****************************************************************************

Do it yourself.  If you forget, and it gets wiped, tough luck!

I usually store all my data in one work disk and one backup disk and is
in my possession.

The data is transfered and stored over an ethernet network connecting
several Macintosh computers (over 200), a Silicon Graphics Power station and
several Indigos used as terminals for graphic display. The whole
process of data transferring is done automatically by means of software
that I developed. (I am planning to attend the next Genome Conference,
South Carolina, USA) to talk about this, and to give copies of the
software. The process of creating the sample sheets for the sequencer
(another time consumming task) and the protocol for the technician's real
work, are also done much faster with the software I developed. We have a 2GB disk on a Silicon Graphics computer, running as an Apple
File Server (via CAP). We are using it as a central storage, and as a
working place for the users. They are able to acces the data both from the
Macintosh and from the Silicon Graphics itself, for running analysis
software (like GCG). The whole disk is backed up weekly. At the same time,
we "download" data to smaller disks (the PLI 21 MB Floptical disks are 
doing a good job). The gels image are kept for about a week. Each one 
takes over 20MB ofspace and are rarely used after that period.

As for hardware, the biggest help we have fund are the syquest 88mb external
drives which take cartridges.  The drives run about $450 (US) and the
cartridges are between $90-100 each.  It makes a great way to deal with the
large volume of data.  We have a mac where we edit sequences so all we need
to do is pull the cartridge from the mac connected to the 373 and put it in 
the drive for the other "editing" mac.  We have had no problems at all with 
these drives.

As far as raw data backups, this is virtually impossible with normal disk 
storage space, but really all of the interpretable information is found on 
the 4-color printout of the "peaks", hard copy of this is essential for 
accurate data interpretation.  

We back up nightly. 

I'm using 128 Mb magneto-optical drives for storage.  The drives are under
$1000 and the disks are about $40 each.  This way everyone in the lab can
be responsible for their own data, since each disk will hold about 1000
analyzed sequences.

****************************************************************************
(4) Storability of sequencing reactions.
****************************************************************************

Presumably you mean completed reactions, mixed + dried down, prior to 
reconstitution in EDTA/Formamide.The most I left it for was over a weekend, 
with no apparent lossin signal.

I have not had much good results with storing the sequencing reactions
once they are resuspended in formamide:EDTA buffer.  But, I guess you
should be able to store the sequenced reactions as final ethanol precipi
tate for few weeks.  May be ABI will have a better opinion in this regard,
although I suspect they will say that it cannot be stored.  If that is
the case, I wonder how is it that they store the reaction standards that
come with many kits.

The cycle sequencing reactions can be stored frozen immediately after the 
reactions are cycled for several days without noticable change. Just take 
the tube rack out of the PECetus 9600 and put it directly into a -20 deg. 
freezer. BUT once you pool and add the loading mix, forget storing them. 
They go off noticable in 24 hours if frozen with theloading mix (don't 
really know why but they do).

****************************************************************************
(5) Prefered chemistries; Taq vs Sequenase, dye primer vs dye terminator 
****************************************************************************

I used Taq and primers.  If you need to use terminators much (in other 
words, are manifacturing your own non-dye labeled primers), then you should 
have bought the pharmacia machine.  The ABI machine's advantages are really 
with the dye-primer system.  I found that I would routinely get about an 
80% success rate with primers (failures due to impure sample - most often 
RNA contamination -or non-accurate mixing of the post reaction samples 
usually) whereas that dropped to about 40% with the terminators.

As of now, Taq appears to be the superior system for sequencing double 
stranded templates.  May be if you have a large M13 as a single stranded 
prep and ready to primer walk, it may be beneficial to use the sequenase 
system.  I am quite positive that there is going to be further optimization 
regarding the sequanase system and it will be worthwhile and consistent, so 
make sure you have the fifth filter wheel required for sequenase. I hate 
the dye primer because you need to do separate reactions. Conceptually 
since both systems incorporate only one dye label on each molecule, I don't 
think it matters much with respect to sensitivity of detection.  Also, in 
our instance, we have different primers to work with and it becomes 
exhorbitant to use labeled primers for each reaction. May be you have to 
take a look at the whole setup you intend and if you are doing a lot of 
sequencing off strandard primers, it may be worthwhile  to consider using 
the dye primer system.

I like the ABI sequencer because of the variety of chemistries that I can do 
with it.  Which one I use depends on the template I start with.  Most of the 
templates I use are PCR products and dye terminator chemistry is flexible
for these.  I have also done some dye primer reactions on nested deletion 
sets of cloned DNA.  The results from these reactions are very nice. You 
need to have the proper clone which may be more work than PCR.  The guy at 
the University of Georgia core molecular biology lab (the kind of core lab 
where you give him the sample and he gives you back the data with a bill)  
told me that about 95% of his submissions prefer dye terminator sequencing 
of PCR products (but that may be because that is the method that they are 
most familiar with). These are all Taq sequencing reactions.  I do not have 
Sequenase capablityon my machine.  Other people that I have talked to about 
it are not sure yet.

We have only used the Taq cycle sequencing so I can not comment on the
chemistry differences.  The use of dye primer or terminator often depends on
the projects, both give good results.

We use dye terminator chemistry and Taq cycle sequencing  for direct 
sequencing of ds PCR products with good results.  

Lastly, remember "simplest is best", get ss (m13) templates to work first
and buy the kit from ABI, then get ds (pUC or other plasmid) templates to
work second, again using the ABI kit.  Then try the terminator kits from
ABI with the above template preparations that you know work well with
the fluorescent-labeled primers.  Once all that's been done and standardized,
then give PCR-product based sequencing templates a try.  Again, the
philosophy here is to try the simplest templates first and then move on to 
templatesthat are more difficult to isolate at a sequence quality purity.
Note on kits:  I hate kits and usually don't advocate buying them BUT ABI 
has a vested interest in getting their instruments to work and work well 
and is the sole source for these kits which do work quite well.You can make 
your own reagents, but for a site with on 373A, the bother is not worth it.

Simplest is Taq cycle sequencing with dye primers on ss templates.  Our 
experiences with the new 5 filter wheel and the Sequenase terminators is 
excellent.  Much better than with the Taq terminators.

All I've ever done is Taq cycle sequencing of PCR products.  It works fine
for me.  The keys are getting enough template, and getting clean template.
 I always gel purify (by spinning the gel chunk through filter paper). 
A 'typical' 100 ul PCR reaction gives enough template for six or more
sequencing reactions.  Also, I always sequence both strands.  There are
generally a couple of ambiguous bases per sequence, but the ambiguous bases
are always in different positions on the different strands, so it's not a
problem.

****************************************************************************
(6) Level of maintenance cover
****************************************************************************

It costs quite a lot to run this setup.  The sequencing kits run for
about $600 or more and the glass plates run at about $200 a piece.
Also I heard that the Argon (?) laser system in ABI machines has
a life span of about 4000 hours or so.  So be prepared to make
contingency arrangements in case it does go bad in about a year or two.

The first year your machine will be covered completely under warrenty. After
that a maintenance contract is essential.  There is a couple of very
expensive parts that have short lifespans.  Also, people not on contracts 
get put on the bottom of very long priority lists for service.

As for service contracts, I have found it worthwhile so far.  They are very
expensive, but in 2.5 machine years we have had 2 PMT failures, 2 lasers, and
1 power supply go as best as I can remember.  So for us, it has been worth
the cost.

Get the FULL service contract.  Our laser dies at least once a year
and the cost of that alone is worth the FULL (I think they call it GOLD)
service contract.  We get prompt, excellent service from ABI when one
of the 373A's goes to hyperspace.

***************************************************************************
(7) Problems
***************************************************************************

A word of warning here. When our machine was installed (together with a fancy Mac and 
colour printer mentioned above) we were warned that maybe installing
other systems folders on the computer was not a good idea.  Now,
as I'm sure you know, most programs come with their own systems
folder.  Some time after we had got our machine, there was a conference
or something coming up, and someone wanted to do a colour print out.
So, the sequencer one being the only one available, they copied their 
software (including a systems folder) onto the machine.  I have to
say, they also fucked about a bit with things like the printer driver.
However, the point is that after all this was finished, and their
stuff removed, the next time we ran the machine, it crashed in the
middle of the run (*&^%$#&).  We phoned up ABI, but their response was
that since we had put another program on the hard disc (no mention of
a systems folder) they refused to come out and help us (this within
about 2 months of purchase).  Eventually we had to completely reformat
the whole computer - absolutely everything, and set it up for 
the sequencer.  In a way, since I (as I say, a computer iliterate)
was able to do this and get it up and running, it is quite a good
advert for the system.  However, we were a bit pissed off at ABI,
and after that the computer (which easily had the capability) was
not used for any other programs.

Getting good template.  Getting people to RTFM (read the #%$^& manual).

One part that may throw you untill you master it is removing the
PCR primers before doing the sequencing reaction.  We have been using
CENTRI-SEP Columns from Princeton Separations, Inc (Diagnostic Division, P.O. Box 300
Adelphia, N.J. 07710 [908-431-3338]).  These are the columns recommended by
ABI and we settled on them after wasting time with other systems we already
had on hand. 

****************************************************************************
(8) General Comments 
****************************************************************************

Data processing with this instrument does take lot more time than
that is portrayed by the ABI.  We routinely analyze each lane for accuracy
of tracing which is required in most cases unless the run happens to be
very good.  Processing one gel file of 36 sequencing reactions, look at
each lane, call bases, output the sequence, print the graphs and other 
sundry stuff takes almost one days work.  So this makes sense to have
the software installed in multiple MACS so that the analysis can be done
independent of the machine.  Very soon the data that is output by the
machine kind of overwhelms you, unless the work is progressing so fast
that this does not become a bottleneck.  

I used one of these babies when I was in Japan about 6 months ago.
I was running it at least twice a week for at least 6 months (don't
you just HATe genome project sequencing?).  I loved the machine on
the whole - a couple of problems, but more of that later.

I have a ABI sequencer that is not really overwelmed in its use.  However,
I have found a few resources that have helped a lot.  One resource is Brian 
Holloway of the U.S. Centersfor Disease Control in Atlanta, GA.  His phone 
number is (404)639-2413.  (Idon't have an Internet address for him but I 
know he has one.) Brian is a real eager 'techy' who has lots of good info.  
He is in charge of a core lab that has the responsibility for providing 
high tech equipment toresearchers at teh CDC.  He does not provide the 
labor for the equipment.  So he is experienced with numerous people of 
varying expertise using his equipment.He has about a dozen of the ABI
sequencers and about 50 users.

I am looking forward to create an active ABI users group.
I understand that there is such a group, but so far I received no
answer to my request for inclusion. The existence of a known list of
users may help us to share solutions and to anticipate problems.

I have access to an ABI 373A, but up till now have had 
difficulty getting funding for systematic studies using it!  Comments 
like "is it accurate enough" continue despite its proven use in hundreds of 
labs...  Hopefully the use of this technology in systematic studies 
at the British Museum will begin to convince the systematics community 
in the US of the validity of this approach.

I also STRONGLY recommend that each gel run contains a STANDARD
that YOU KNOW works and a second STANDARD template reaction run by
the individual who gives you their samples.  The biggest problem is
garbarge templates prepared by folks who don't really know what
they are doing.  You should create a protocol for ss and ds template
preparations and get everyone to use it or else no guarantees that
the data will be reasonable.

Since you've got the money for the 373A, also buy a PECetus 9600
for the cycle sequencing reactions.  NO OIL required and very reproducable
temperature cycling (disclamer: I don't work for them or have any personal
gain, just a very satisified user).

Since you have purchased only one machine at this time I would recommend 
that you do the data analysis and chromatagram printing on a separate 
computer. That will free up the sequencer so that you will be able to do 
two runs a day, an 8-10 hour run during the day and an overnight run. You 
will just need to remove the Analysis software from the computer that 
will be doing the data collection. When the data collection has finished 
you will need to transfer the gel file to the analysis computer (I do this 
over my apple talk network) and begin Analysis by double-clicking on the 
gel file. Thea nalysis and chromatagram printing will then proceed as usual.

For someone who is already familiar with 'old-fashioned' sequencing, a
couple days training should be enough, in my experience.  

The latest pricesheet I have lists them (plates) at $530 a set.  They're
made of "optically flat, low fluorescence Vycor glass from Corning," 
according to a paper in Gene Analysis Techniques and Applications 9:9-16
(1992).  Has anyone looked into buying some of this glass and having the
local glass shop make a few plates?

The Ericomp Twinblock and Powerblock cyclers with heated lids also don't
require oil (definitely a plus when doing the small-volume sequencing
reactions), and cost about half as much as a Perkin-Elmer 9600.  The Idaho
Technologies Air Thermal Cycler, which does the reactions in capillary
tubes, is very fast, doesn't require oil, and also costs about half as
much as the PE 9600.  Despite initial worries about the halogen lamp in
the Idaho machine fading the dyes, I'm getting excellent sequence using
one.  You just have to remember to add some BSA to the sequencing reaction
to keep the taq from sticking to the glass.  Standard disclaimers, of course.

You can reduce costs by doing 10 ul reactions instead of 20 ul, keeping
the concentrations of everything the same.  There is still plenty of
signal (at least with dye terminators--I've never done dye primers).
I suspect you can also reduce the concentrations, and I'll be doing some
experiments soon; anyone out there know how little of the expensive dye
terminators we can get away with?

***************************************************************************
***************************************************************************
Thank you one and all. Happy sequencing, be-it manual or auto.

Cheers

DAJ
David A. Johnston
Dept of Zoology, The Natural History Museum, Cromwell Road,
South Kensington, London SW7 5DB.
(tel 071 9389297, fax 071 9388754, email daj at nhm.ic.ac.uk)



More information about the Methods mailing list