Wednesday, November 9, 2011

How did I become a graphic designer?

The public perception seems to be that grad students and other academic types are smart. You know, abstractly brainy, maybe good at math... What no one seems to realize is that in science not only are we supposed to be able to comprehend abstract concepts, to be successful we need to:

  • Write well
  • Be good at public speaking and presentation
  • Fix equipment (in my area, this means I can duct tape stuff, fix plumbing, wire things, solder, super-glue, etc.)
  • Create what you need from scratch out of stuff you find around the lab or buy cheap
  • Negotiate with product vendors to get good quotes on stuff
  • Be proficient with data analysis and graphing software
  • Be a graphic designer
It's this last one that trips me up (ok, that and the fact that I have wretched eye-hand coordination). I used to complain about this in my experimental lab, where all I had to generate was graphs. I love a good graph, it's always great when you graph something and suddenly a pattern is clear. But then you have to get these things ready for publication/presentation. I'd make a (simply dazzling) figure, and my PI would say, "Hmmm, maybe increase the axis line weights, add minor tick marks, change the color of the one line to a darker red and don't have little serifs on the scale bar." I would do all that, he'd look again and say, "What if we used open circles instead of filled circles? Let's show the correlation with a dashed line, not a solid one. Also, how about swapping panel C with panel F?" And so on.

But little did I know - it could be so much worse. Dun dun dun.

I'm getting preparing my first manuscript in the computational lab that I've been working in. I've got some really cool structural simulation data, and now I have to find the perfect way to display each aspect with 2D images. Really, it's all best shown with movies, but you can't publish a movie. So here I am in my program (called VMD - visual molecular dynamics) sitting on a pile of choices.

  • Do I use depth cuing? What perspective? Which frame of the movie? 
  • Then there are all the ways of showing a protein structure. Do I want my helix to look like a cartoon twisty? A cylinder? van der Waals radii of all atoms? Sticks? Ball and stick? Only show the protein surface? Implausible things, like licorice and paper chains? 
  • Then I can choose how to color: by atom name, residue name, secondary structure, subunit, charge, conformation... 
  • I also have a choice of "material" which is what the representation I've chosen is "made of" in the program. These choices have some obvious ones, like opaque, transparent, shiny and also some weirder ones like ghost, glass, brushed metal and chalk
  • Don't like the preset red? You can adjust it with the color manipulator. How opaque is opaque? There's a slider for that too!
Worse still, I'm often superimposing structures, or showing ligand molecules with the structure, so I may have two cartoon helices superimposed in different transparent colors with a ghostly surface representation of the rest of the protein and the drug molecule colored by atom name in licorice representation.

All of these parameters take ages to tweak so that my awesome discoveries are shown looking their best. I will never again criticize structure figures in any paper. And my advice is, if you're looking at a structure paper and they have a supplementary movie, save yourself some squinting and download it already!

No comments:

Post a Comment