[BioC] specifying hyperlinks on the plot

Jane Fridlyand janef@stat.berkeley.edu
Sun, 3 Nov 2002 15:40:57 -0800 (PST)


currently i am specifying clone names with a very small cex size and
passing images on the biologists in the pdf format who then open it up in
the Photoshop which i guess has a very good ability to zoom in (maybe
adobe acrobat does too: i just don't have the full version). In any case
focusing the mouse on the right clone names is not a problem for them. But
i guess Vincent's concern is the file size with hyperlinks encoded for
each clone name, is it right? Sparcing links across the genome would also
be an option of course. What would be a reasonable number of hyperlinks
per image?

Jane







**************************************************************************************
Jane Fridlyand, Postdoctoral Scientist
UCSF Cancer Center, Box 0128 San Francisco, CA 94143-0128
Office: Room N412 Tel: (415)514-3302 Fax: (415)502-3179
**************************************************************************************

On Sun, 3 Nov 2002, Vincent Carey 525-2265 wrote:

>
> On Sat, 2 Nov 2002, Byron Ellis wrote:
>
> >
> > On Saturday, November 2, 2002, at 01:55  PM, A.J. Rossini wrote:
> >
> > >
> > >     vincent> the short answer is no.  a related question is why
> > > restrict to pdf
> > >     vincent> or postscript?  for x11 graphics, the locator function
> > > can be used
> > >     vincent> to get information about mouse position.
> > >
> > > Restriction may be the wrong word, but the general idea is solid, I
> > > think (i.e. mailing a hyperlinked PDF file to collaborators, for
> > > example).
> > >
> >
> > Even beyond this, PDF also supports embedded Javascript (for better or
> > for worse) that could let you do potentially interesting things. Being
> > able to generate active PDFs is definitely desirable (for posting or
> > mailing to people who neither know nor care about R :-)). I don't see
> > the current device driver model handling this very well.
>
> ok, i am warming up to this idea, but i suspect that the density
> of hyperlinks to be handled by existing designs would be pretty
> light (at most a few links per line of narrative text or per embedded image).
> what jane is describing sounds very dense, and the quantity of
> metadata in a chip image could be substantial even if only one
> sort of link resolution were required ([x,y] to locuslink, e.g.).
> one would want to do the mapping from the geometry of the document
> (image) to the genomic annotation outside the document -- so i
> would assume that the document and a database would be involved
> at a minimum.
>
> i made a cursory search on active pdf and a check of the adobe
> site.  i don't see anything immediately available for prototyping.
> one approach is to generate the pdf image in R and then apply
> some adobe tool to introduce a hyperlink.  i would assume distiller
> could do this but i do not have access.  if this worked we would
> then consider how best links could be introduced en masse.
>
>