Input requested on best DNA Sequencer

Peter Gegenheimer peterg at
Mon Jan 9 18:07:16 EST 1995

In <1995Jan3.140237.29429 at>, aarment at (The doctor is IN.) writes:
>       I am preparing to buy a sequencing apparatus and power supply with
>some recently acquired grant monies and would like some input from other users
>out there as to what features and companies I should look at.  Unfortunately,
>funds are limiting, so I have to be somewhat thrifty. :)
>Please post to this newsgroup or directly.  I will compile the replies if there
>is enough interest.  Thanks.
>aarment at

Appended is a comparison of features I've prepared based on my experience.  
Our service lab has recently purchased the LI-COR 4000LS sequencer. The lab
which uses it the most, consistently gets 700 to 1000 readable bases per run. 
A colleague at another university just told me that, in the course of 
evaluating sequencers, they submitted identical samples to ABI, LI-COR, and 
Pharmacia. ABI and Pharmacia read about 300 bases; LI-COR read 800 bases.

Folks, following is a comparison between the three major brands of automated 
DNA sequencer currently on the market, made by ABI, Pharmacia, and LI-COR.  I 
prepared this when we purchased a LI-COR sequencer for our service lab this 
summer (1994).  The result s of our investigations concluded that ABI or 
LI-COR were the leaders, and that for our needs (a low-budget, multi-user 
service lab), the LI-COR was clearly preferable.  The following information 
will help put this into perspective.  

First, only the ABI instrument can run all four sequencing reactions in a 
single lane.  That means that you CAN get four times as many sequences as the 
other machines.  (The four reactions don't run in the same ladder, however, so 
their electrophoretic mobility must be computer-corrected.)  On the other 
hand, the ABI will read only about 450-550 bases per run, whereas the LI-COR 
will routinely generate 1100 bases of sequence, of which 700 to 900 bases are 
accurately readable with operator assistance.  (A skilled operator can read 
1000 bases at a run.)  The C. elegans genome project uses ABI's for routine 
sequencing, and Li-Cor's to close gaps.  Even Lee Hood's lab is starting to 
use Li-Cor for that reason.  I have not heard anything good about the 
Pharmacia ALF machine, and I was not impressed by its technical specs nor by 
the presentation from the sales rep.  I have been told that the Pharmacia 
machine is not identical to the EMBL instrument upon which its design is 

  Second, from a technical perspective, the Li-Cor instrument has the most 
state-of-the-art features.  Although other designs can be adapted for long 
reads, Li-Cor has patented the infrared detection technology which gives their 
detector a lifetime 10 times greater than that of the others.  

  Third, when we got our Li-Cor, I didn't expect it to yield good sequence 
without several week's tuning-up.  I was quite surprised, then, that the same 
day it was unboxed and set up, and reactions loaded by our technicians who had 
never handled DNA before, we generated over 1000 bases of sequence.  650-700 
bases were readable by fully-automatic base-calling (at several hundred bases 
per min), and 915 bases were readable with operator assist beyond 700 bases.  
Our service lab has found Li-Cor to be very helpful with all stages of use, 
from installation to both the chemistry and machine use.  

  Disclaimer:  I have no formal connection with Li-Cor.  A former graduate 
student of mine is now (after an academic post-doc) an Applications Scientist 
in Li-Cor's Biotechnology division.  However, I also have high regard for 
ABI's scientific and technical abilities.  

  What follows are the specs we presented to the State Purchasing Department 
in Summer 1994.  


  1. Must be capable of analyzing DNA sequencing reactions in which the 
product DNA fragments are labeled with a nonradioactive molecule such as a 
fluorescent tag.  DNA fragments must be detected in real time without 
disassembly of the electrophoresis gel.  
  * JUSTIFICATION:  An instrument which uses radioisotopes is impractical Ä 
for safety and regulatory reasons Ä for daily use in the Core Facility.  
Nonradioactive analysis methods inherently provide greater resolution, i.e., 
can read greater lengths of DNA, which is a major reason for acquiring an 
automated system.  

  2. In totally unassisted mode, must be certified and documented to read over 
500 bases of DNA with greater than 99% accuracy.  In operator-assisted 
semiautomatic mode, must be certified and documented to read over 900 bases, 
and up to 1200 bases, of DNA with greater than 99% accuracy.  
  * JUSTIFICATION:  the primary reasons for acquiring an automated sequencer 
are to read more bases at once, and with greater accuracy, than manual 
methods.  The participating users absolutely require the ability to read 800 
to 1000 bases in fully automatic mode under ideal conditions, so that under 
suboptimal conditions we can still read over 500 bases.  The ability to read 
1000 to 1200 bases in operator-assisted mode is essential to the economical 
operation of the instrument.  That is, the more bases that can be read at one 
time (in one "run"), the fewer separate runs must be made to cover a large 
region of DNA.  Since the cost of a sequence determination is based on the 
number of separate runs, not on the number of bases read, it is important to 
use as few runs as possible. 
  * MEETS SPECS:  LI-COR:  YES Ä rated and tested to read at least 800 bases 
auto; ca.  1000 bases semi-auto.  ABI:  no Ä tested to read 550 bases.  PHARM:  
no Ä rated to read 450 to 500 bases.  

  3. a) Must be demonstrated to accurately sequence single- or double-stranded 
DNA and to sequence directly PCR products.  b) Must permit the use of either 
fluorescently-labeled primers or fluorescently-labeled nucleotides.  c) Must 
accommodate DNA sequencing reactions produced by Sequenase and 
cycle-sequencing methods.  
  * JUSTIFICATION:  a) A major group of users will be systematics researchers 
who can supply only PCR-amplified DNA samples made with a large number of 
different primers.  b) The sequencer should be able to use the researchers' 
own primers, which are not fluorescent, and incorporate fluorescent "tag" 
using fluorescently-labeled nucleotides.  c) Other users will be supplying 
DNAs which are single- or double-stranded, and the sequencing service must be 
able to handle all these samples.  
    [Note: LI-COR is reworking their internal labeling protocol, so that for
the moment they guarantee results only with led-labeled primers.]

  4. a) Gel electrophoresis chamber must accommodate gels up to 66 cm in 
height, for reading sequences farther than 700 bases.  b) Must be able to 
determine short regions of DNA sequence in 1-2 hr on a short gel as described 
above.  c) Must accommodate smaller electrophoresis gels between 18 and 25 cm 
used for analysis of RNA and proteins.  
  * JUSTIFICATION:  a) The required ability to sequence long regions of DNA 
(up to 1200 bp), as set forth in (2), can only be accomplished with long gel 
plates.  b) However, some users will need only short sequence readings.  This 
can be done in several ways, but the most economical is to use shorter 
electrophoresis gels.  These short runs will also be ideal for teaching 
purposes.  c) In addition, several users will be running DNA and RNA analyses 
called "fragment length analyses".  These are performed on shorter gels.  
  * MEETS SPECS:  LI-COR:  YES Ä accommodates gels up to 66 cm without 
modification.  ABI:  no Ä maximum is 34 cm.  PHARM:  no Ä maximum is ca.  45 
cm (cannot exceed 50 cm).  

  5. a) Gels must be temperature-controlled with an integral heating device 
for precise heating during DNA sequencing runs.  b) Electrophoretic field must 
be user-settable to continuous or inverse-field modes for maximum resolution 
of large DNA fragments.  
  * JUSTIFICATION:  a) Optimal resolution ("detail") in a DNA sequencing run, 
and in RNA fragment length analysis, and certain types of DNA fragment length 
analysis, requires that the gel be heated to 40ø-60øC.  b) Certain types of 
fragment-length analyses require good separation of large fragments which can 
only be accomplished by field-inversion gel electrophoresis ("FIGE"). 
  * MEETS SPECS:  LI-COR:  a) YES, b) YES.  ABI:  a) no Ä uses passive 
heat-transfer plate, no heater; b) no.  PHARM:  a) YES, b) no.  

  6. Machine as configured must be able to collect data suitable for DNA 
fragment analyses including STR (short tandem repeat), SSCP, RAPD, and 
comparable protocols.  
  * JUSTIFICATION:  Several important groups of users, ones upon whose support 
the funding for this machine was predicated, are doing population genetic 
studies which use STR, SSCP, and RAPD analyses.  They will need to be able to 
run their samples on the sequencer, and then move the data to their own 
computers for detailed analysis.  
  * MEETS SPECS:  LI-COR:  YES.  ABI:  no, requires $27,000 software upgrade.  

  7. Must be usable with gel electrophoresis plates made of standard soda-lime 
glass.  Replacement cost of longest glass plates must not be over $100.  To 
accomplish this, DNA detection unit must employ a laser which excites at a 
wavelength, such as the near infrared, where interference by plain glass is 
  * JUSTIFICATION:  The Core Facility does not have a large budget for 
replacement and repair, and can ill-afford any "downtime" resulting from 
equipment failure.  The glass electrophoresis plates are the component most 
likely to break during normal use.  Detection systems based on ultraviolet 
(UV) lasers require the use of borosilicate or quartz plates, available only 
from the system vendor at a cost of ca.  $300.  In contrast, detection systems 
based on infrared (IR) lasers can use plain glass (float glass or soda-lime 
glass) plates, which are flat to within 0.01 mm but are inexpensive ($50 to 
$80 from vendor in the largest sizes).  
  * MEETS SPECS:  LI-COR:  YES, uses soda lime plates costing $50-80.  Laser 
operates at 785 nm.  ABI:  no, uses borosilicate plates costing $300.  Laser 
operates at 488 or 514 nm.  PHARM:  no, uses a pair of borosilicate plates 
costing $155 (front plate) and $900 (rear plate).  Laser operates at 488 nm.  

  8. Detection assembly (laser and detector) must be composed 100% of 
solid-state components for long-term reliability.  The laser must be solid 
state (diode or equivalent); the detector must be solid-state.  
  * JUSTIFICATION:  Conventional gas lasers inherently have a limited lifetime 
under continuous operation.  Under the usage schedule projected for the Core 
Facility, a gas laser would have to be replaced every three years or so.  A 
solid-state (diode-type) laser, on the other had, has an expected life of 
100,000 hours, or 30 years of Core Facility operation.  Similarly, solid-state 
light detectors have a much greater life expectancy than conventional 
photomultiplier tubes. 
  * MEETS SPECS:  LI-COR:  YES, YES Ä laser is solid-state, rated for 100,000 
hours and warranted for 5 yr.  Detectors are photodiodes.  ABI:  no, no Ä 
laser is argon gas; detector is photomultiplier tube.  PHARM:  no, YES Ä laser 
is argon gas; detectors are photodiodes.  

  9. Detection assembly must be focusable to permit use of gels ranging 
between 0.1 and 1.0 mm thick for analytical and preparative work. 
  * JUSTIFICATION:  Ultra-thin gels, 0.1 mm thick, are needed for maximum 
accuracy and resolution (greater than 1000 bases) and for high-speed runs.  On 
the other hand, thicker gels (1.0 mm and up) are needed for preparative RNA 
and DNA work, and analysis of larger amounts of material as required by some 
of the users. 
  * MEETS SPECS:  LI-COR:  YES Ä detection assembly (laser etc.)  incorporates 
a miniature confocal microscope, adjusts in two dimensions, and is 
auto-focussing.  ABI:  no.  PHARM:  no.  

  10.  a) Must come equipped with a hard-copy device (printer) which allows 
the operator to make routine hard-copy printouts of the raw electrophoresis 
data.  b) The hard-copy must be a two-dimensional facsimile of the gel itself 
(e.g., an autoradiograph-like image or comparable) and c) must be completely 
readable in black-and-white. 
  * JUSTIFICATION:  a) Most users have told me they will not accept the 
computer's reading of the data unless a printout of the raw data is included 
for inspection.  b) It is impossible to determine the quality of the data, in 
particular to ascertain the absence of "smiling" and other gel artifacts
unless a gel facsimile can be analyzed.  Furthermore, for fragment analyses, a 
conventional photograph-like gel facsimile image is the easiest and best way 
to present and interpret the data.  c) For publication of the raw data, or 
transmission to other researchers, a black and white image is easier and 
cheaper than a color image.  
  * MEETS SPECS:  LI-COR:  a) YES.  b and c) YES.  ABI:  a) no, printer costs 
$4200 extra.  b) no, hard-copy is one-dimensional computer-adjusted 
chromatogram of each lane.  c) no, printout is in color, cannot be read in 
black-and-white.  PHARM:  a) YES.  b) no (same as ABI).  c) no (same as ABI).  

  11.  a) Control unit and computer must be capable of simultaneously 
acquiring data from one or more electrophoresis units in real time, analyzing 
stored data, and transmitting data over a network to end-users.  b) A 
preemptive, multitasking operating system (IBM OS/2 2.1 or equivalent) is
required to accomplish this. 
  * JUSTIFICATION:  a) The instrument will be operated in the Core Facility 
and will serve multiple users from different labs.  The more-sophisticated 
users will want to be present during the data interpretation phase (called 
"operator-assisted" above) to increase the amount of usable data.  In 
addition, since data files are often too large to save on a floppy disk, they 
must be transmitted to the user via Ethernet connection.  It is thus 
imperative that data interpretation and data transmission be able to occur 
simultaneously with data collection.  b) Simultaneous operation of multiple 
procedures without "losing" data can be accomplished only by a computer 
operating system having "preemptive multitasking".  The OS/2 operating system 
and UNIX are two such systems.  We prefer OS/2 for compatibility with 
existing software, and because the operators and users are more familiar with 
it, and the P.I. can provide technical support.  
  * MEETS SPECS:  LI-COR:  a,b) YES Ä Uses IBM's OS/2.  Control unit and 
computer can actually acquire data from three electrophoresis units 
simultaneously.  ABI:  a,b) no Ä uses Macintosh "System 7" operating system 
which is only partially ("cooperatively") multitasking.  PHARM:  a) not 
described.  b) YES Ä uses IBM's OS/2.  

  12.  a) The system as provided must allow for replication of the 
base-calling software onto one or more remote workstations, and b) the raw 
image data from the instrument must be transmissible via the campus network to 
these workstations.  
  * JUSTIFICATION:  a) the technicians in the service lab do not have time to 
provide more than basic machine operation with automated base-calling; 
operator-assisted base-reading is best done by the individual users at their 
own locations and on their own time.  
  * MEETS SPECS:  LI-COR:  YES, YES -a) basic software license allows ca.  4 
remote copies at no extra cost, b) the control computer comes equipped with 
network card and TCP/IP software.  ABI:  no, not stated.  PHARM:  no, not 

WARRANTY: Finally, Quoted price must include on-site delivery and training.
  Minimum warranty must be three years of on-site service on both 
electrophoresis unit, power supply, system controller, and computer.  
  * JUSTIFICATION:  The instrument will be operated heavily to serve many 
users.  Any downtime will adversely affect the Core Facility's ability to 
serve its users.  More importantly, the Core Facility does not have a large 
budget for instrument repair.  A multi-year, on-site service contract will 
greatly smooth operations.  
  * MEETS SPECS:  LI-COR:  a) YES.  b) YES/almost Ä 2 years on-site service on 
all components except 3 years on-site service on computer and 5 years service 
on laser/detector assembly (on-site in first year only).  ABI:  a) YES.  b) no 
Ä one year only.  PH ARM:  a) YES.  b) no Ä one year only; 3 months on 


  LI-COR:  One each Part Numbers 4000LS; 9940-079; and 4000-52.  $79,380.00 
(Quote of 4/29/94); complete with system controller and power supply; 
multitasking computer; one electrophoresis unit with two gel stands ("Gel 
Apparatus"), one 66 cm and one of our choice; two sets of gel plates, 
spacers, and combs; two printers (one text, one for gel images); one reagent 
kit and one set of DNA primers.  2/3/5 year warranty as described above.  

  Perkin-Elmer/Applied Biosystems:  Quotation #259094 (4/27/94).  One each 
Model #373A-18; $85,000.00 complete with system controller and power supply; 
non-multitasking computer; electrophoresis unit with one short gel stand (34 
cm), two sets of gel plates, spacers, and combs; no printer; one reagent kit.  
One year warranty.  NOTE:  This system can do sequencing on a smaller number 
of samples than the regular ABI #373A, it cannot do any fragment analysis, 
etc.  Upgrade to a complete system costs $27,000.  ABI also sells a 'long-gel' 
system claimed to sequence 500 to 700 bases, for about $120,000, but I have 
herd nothing about its abilities in the field.  

  Pharmacia Biotech:  Quotation # 6744-042794-00 (4/27/94).  One each Part 
number 18-1004-S1; $90,000.00 complete with system controller and power 
supply; multitasking computer; one electrophoresis unit with one gel stand 
("upper and lower buffer reservoirs") , two sets of gel plates, spacers, and 
combs; one plotter (for one-dimensional tracings only); no reagent kit.  One 
year warranty.  

|  Peter Gegenheimer            |  pgegen at             |
|  Departments of Biochemistry  |  voice: 913-864-3939                   |
|    and of Botany              |                                        |
|  University of Kansas         |  FAX  : 913-864-5321                   |
|  2045 Haworth Hall            | "The sleep of reason produces          |
|  Lawrence  KS  66045-2106     |  monsters."              Goya          |

More information about the Methods mailing list