Phred Quality analysis

Osborne, Brian Brian.Osborne at
Tue Jul 17 09:17:41 EST 2001


I can't speak to the 3700, the last work I was involved in with DNA
sequencing used 377's and 373's, but we would typically define
"high-quality" as a sequence having an average Phred quality of 11 over a
window of 15 nucleotides. So the perl script would start at the ends of each
read trimming sequence that didn't meet this criterion until it reached high
quality sequence. These values were determined empirically, by examining
many reads and noting the Phred scores of what scientists considered legible
and illegible sequence.

I notice that Mark talks about phred's trimming feature, we didn't use this

Brian O.

 -----Original Message-----
From: 	Mark T Jung [mailto:MARK.T.JUNG at] 
Sent:	Tuesday, July 17, 2001 6:40 AM
To:	autoseq at
Subject:	Re: Phred Quality analysis


     We have been using phred for about a year now to QC our sequencing
results and have found this to to be
very useful.  My impression is that most others would agree with this.  We
did a direct comparison a few months
ago with Trace Tuner (Paracel) and found we liked the phred quality
assignments better.  Paracel claims a 10%
greater read length on 3700 data; we could not confirm this on our 3700
data.  We use phred's trimming feature to
trim our data.  E-Mail me directly if you want specifics regarding how we
do the analysis.

Mark Jung

Tony <darthvaderscat at> on 07/12/2001 10:06:25 AM

To:   autoseq at
cc:    (bcc: Mark T Jung/AE/DuPont)
Subject:  Phred Quality analysis

Has anyone out there has any sucess storied with long term monetering
of sequence quality using the phred<20 scores for the 3700?
I am currently trying to get a quality system based on phred scores
internally validated and I would like to hear from people who have
done the same thing and how they got on. Amongst my many problems are
assigning cut-offs.
Thankyou in advance



More information about the Autoseq mailing list