Re: dbEST

Christopher_D_Southan@SBPHRD.com
Thu, 16 Sep 1999 08:49:41 +0100

Matto, the only reliable way to determine the frame of an EST is by a good
similarity match, in translation mode, against a protein. You can do this
by a TBLASTN of your protein against dbEST, or, by BLASTX of the EST
against the nr protein database. Although some of the recent rodent reads
are of remarkable quality the error rate in human ESTs is usually too high
to pick out the frame by looking for the least stops from a regular
translation program. Having said that there is a new tool announced on
the EST machine site (www.tigem.it/ESTmachine2.html) called ESTScan that
claims to pick the correct frame by hexanucleotide frequencies, but I have
not tested this (has anyone else out there?). The deposited reads in dbEST
are bi-directional. Although Image consortium reads are labelled 5' or 3'
a proportion of these are "flipped" i.e. they read in the opposite
direction. Remember a high proportion of ESTs are primed off 3' UTR and
will therefore give no frame matches. Another approach is to assemble or
contig the ESTs which may collapse down to give a clearer idea of the
correct frame. The TIGR human/mouse/rat assemblies are good examples but
have long update intervals. The EST machine web site listed above gives a
good selection of links for you to explore.

Yours, Chris Southan, SB UK

villain@uab.edu on 15-Sep-1999 15:26

Please respond to Villain@uab.edu

To: abrf
cc: (bcc: Christopher D Southan/RES/PHRD/SB_PLC)
Subject: dbEST

I have a question regarding the dbEST.( Expressed Sequence Tags).
The DNA sequence present in this database are stretch of mRNA. There is
any way to determine the reading frame (the logical way I thought was
the reading frame should be the one without stop codons, but since the
stop are only 3 out of 64, I think the probability to get three complete
translations are too high). And the other question is, are the sequence
always deposited as sense DNA, meaning when you create this DNA, do you
know which is the sense strand.
Thank you for any answer.
Matteo

--
Matteo Villain                   E-mail : villain@uab.edu
Research Associate               Phone  : 205 934 3032
University of Alabama            Fax    : 205 934 1446
Birmingham
35294, USA