Re. General stuff.

Jim Jim
Wed Jun 18 16:10:49 EST 1997


>3. I recently had an XL upgrade to one of my 377's (at no small cost) 
>why is it that in about 30 gels (at 48 lanes each) I have to re-track EVERY 
LANE ON 
>EVERY GEL? Talk about a weak point. Can someone at PE-ABI write some decent 
>tracking software? PLEASE?? My other 377 uses the older software as gets it 
correct about 
>90% of the time. Not bad for an 'old timer'!!!!!!!

>Oh, and before I forget (let's call this point 4...) will everyone use 
>the (possibly cheaper??) Amersham dye terminators if they don't want to/can't 
use the 
>'new' PE-ABI dye kits? How many poeple already use them? I found NO 
difference 
>whatsoever between the kits ('old' ABI and Amersham) when used. So will 
people go 
>for a saving or better performance?

>A bit of discussion if you please!

Lane tracking software (and sequencing software in general) is a field 
unto itself.
There is some interesting history behind different packages and 
approaches to DNA analysis
software.  ABI is hardware limited in a couple areas, specifically to the 
number of pixels
comprising a gel image. There are some environment issues such as the Mac 
OS operating system,
and then there are the lane finding algorithms that they choose to use, 
including how many 
scan lines to search over an image to establish a neighborhood around a 
pixel to search out
a lane.  I'm not sure how the hardware upgrade affected the software; PE 
did increase the number
of pixels per scan, but I don't think it was that substantial an 
increase.  Now you're packing
more lanes into a gel image [probably] using the same approach you used 
for a fewer number of 
lanes.  Like I said, a field unto itself... If you're really interested 
you might want to try and find 
a paper on the internet by Clark Tibbetts that details some of the ABI 
hardware limitations and how to
overcome them. PE/ABI has a guy named Jim Bowlby working in their 
programming group;  I've had some
good discussions with him in years past about lane finding algorithms and 
basecalling in general. He
might be able to give you a more detailed overview of their lane finding 
method.

You also might try looking at the GetLanes package out of Washington 
University in St. Louis. They've 
done a good job designing a better lane-finder which works with the Phred 
basecaller out of Phil Green's 
group at the University of Washington.

We'ver found that the Amersham kits work a little better for us on our 
device, and the 2-for-1 sale
Amersham is having this month is a bargain.  FS chemistry works good 
enough for most of our sequencing
needs.  We recently sequenced a cosmid using both FS and ET and got about 
the same accuracy and average
confidence per base for our basecaller.  For some examples of lane 
extraction and raw data from our device,
pull up our web page which should be up this Friday (www.genesys-tech.com)

Jim Golden, Ph.D.
Director, Bioinformatics
GeneSys Technologies, Inc.
9757 Wilkinson Rd.
Mazomanie, WI  53560



More information about the Autoseq mailing list