In article <jmack-190494231006 at jmack.pipeline.com>, jmack at pipeline.com
(John Mack) wrote:
> The quest for complete automation is noble, but where does one stop? For
> example, how many high-tech people-years are necessary to create a
> completely automatic (whatever that is) algorithm as opposed to a nearly
> automatic one that requires some low-tech human intervention? It seems that
> perfect automation is impossible UNLESS the gel setup IS standardized. Even
> then there will always be those times when you step just outside the
> envelope and wham! the software chokes!
I should have clarified things a bit - the solution I'm looking for does
not need to be "magic", but it should not arbitrarily restrict you to
having, say, exactly 32 lanes, or a certain lane width, or what have you. I
agree, there is probably no one package that can automatically do every
possible gel on the planet with no hint as to what it might expect in the
image - but one can imagine something that has user-settable parameters
that could allow it to operate on just about anything, given the right
configuration.
In any case, if we are going to get anywhere with large-scale sequencing,
human intervention in the lane tracking process is definitely out.
Eliminating this is the major goal of my current project.
-Anthony
