annotating EST sequence information with BLAST search output

Sean Eddy eddy at wrasse.wustl.edu
Mon Feb 9 08:46:31 EST 1998


francis at BORDUAS.NLM.NIH.GOV (Francis Ouellette) writes:
> Why would one one to annotate any record with information
> which will be out of date the day it is printed?
> 
> I think we would want to discourage such annotations.

That's a valid point. But on the other hand, in many cases, the
annotation is useful, and much of the information does not go out of
date.

For example, if an EST has a 100% identity hit to a known gene, that
information is not going to change. 

A less clear case, but also useful: if an EST shows a blastx hit
against a well-known protein family (say, a transcription factor
family), that's also useful information, even if we don't know yet
exactly what gene the EST corresponds to.

> If you want to see what the record is related to, see the
> blast similarity which are computed in dbEST or in the related
> sequences to a record present in Entrez.  Loooking at 
> annotations which will never be updated is missleading,
> and uninformative.

That's too strong a position to take, I think. Scanning EST
description lines is far, far faster than clicking through to see a
whole dbEST record for each EST. Description lines should contain our
"best effort" to give a short description of the EST.

It's a user education problem to make sure people realize that those
description lines are bound to be outdated, and that they may be
misleading. But "uninformative" they're not, in my opinion.

- Sean





More information about the Bio-soft mailing list