Carrie Sougnez wrote:
> Hello Grace,
> I am from the MIT/Whitehead Institute Genome Center. We also run 8 plates
> a day (really 2 - 384well plates/day) on the 3700. We run 123 instruments at
> a time. We too see a wide range of high and low quality data on the 3700.
> We can usually attribute low quality data to a internal problem in the
> instrument. For example plumbing, camera, and laser alignment problems can
> all cause low quality data. What types of maintenance protocols are you using
> to keep your instruments performing?
>> We also see discrepancy in data quality between the left and right side of the
> array (a difference of 100 phred 20 bases from left to right). This
> correlates to the signal intensity from left to right, but there is no fix for
> this yet... we have had limited success in changing the temperature of the
> cuvette during run modules.
>> As for your resusupension solution, you may want to try 0.5mM EDTA. We tested
> water, formamide, 12.5% pyrillidinone, and EDTA. The 0.5mM EDTA gives us the
> longest read lengths.
>> Carrie Sougnez
> Coordinator, Sequence Detection
> Whitehead Institute Center for Genome Research
> [...]
Carrie,
Thanks for posting your comments here.
By the way, what % of your machines have you switched to version 1.1 data
collection software? I ask because Kevin McKernan posted Jan. 9th [1] saying at
Whitehead they were limiting the number of machines converted to v. 1.1 because of
various problems. I was wondering if those problems had been cleared up or not.
Also it might explain your increase in read length with 0.5 mM EDTA if you were
not using foil piercing (which was implemented in v. 1.1). I have been getting
very good results with v 1.1 [2], but only have a single 3700--so am not a
representative sample.
Phillip San Miguel
Purdue Genomics Core Facility
[1] See http://bionet.hgmp.mrc.ac.uk/hypermail/autoseq/autoseq.200001/0003.html
[2] See http://bionet.hgmp.mrc.ac.uk/hypermail/autoseq/autoseq.200002/0019.html
---
