Скачать или смотреть The essence of data visualization in bioinformatics (Webinar)

Presenter: Martin Krzywinski, Canada's Michael Smith Genome Sciences Center.

Progress in science requires effective communication—do not leave your audience confused, bored, or uninspired. Using critique by redesign and drawing from published examples in bioinformatics and molecular biology, I will show you how small changes to critical elements can turn a muddled figure into one that is clear and concise. You will learn how to discover, evaluate and create structure in visual explanations by paying attention to the building blocks: salience, continuity, color, type, order, Gestalt and layout. The presentation does not assume familiarity with specific software or specialized bioinformatics knowledge.

MARTIN'S ANSWERS TO AUDIENCE QUESTIONS:
1. What is your favorite software for data visualization, plotting of graphs and generating tables?

I use R for general things, Gephi for networks and custom scripts (Python or Perl) for complicated scenarios that I'm not able to figure out in R.

I used to use OriginPro from OriginLab. It's a very competent scientific plotting suite if you like to be guided by an interface.

2. what software do you recommend for figure creation?

Illustrator or Affinity Designer.

3. Is there any guidelines (e.g. website, reference…) you could recommend?

Points of View
http://mkweb.bcgsc.ca/pointsofview/

Elements of Typographic Style
https://en.wikipedia.org/wiki/The_Ele...

4. Your thoughts on violin plots? Great for showing possible subpopulations of results (which box and whisker plots don't) or kill them with fire?

If the distributions are bimodal or have a structure that is important to the message then violin plots can be helpful. Keep in mind that the convolution can smooth the distribution and hide patterns, too. Consider showing the distribution without any smoothing beside the box plot.

5. I’ve had a reviewer who was colour blind and therefore I had to change my figure colours to colour-blind friendly colours. Colour-wise in figures for papers, should we use colour-blind friendly colours by default? How can we best show contrast?

Palettes for colorblindness
http://mkweb.bcgsc.ca/colorblind/

Color Oracle to simulate color blindness
https://colororacle.org/

6. What additional packages in R do you use besides ggplot?

lattice, rcolorbrewer, plotly

7. Should we always use sans serif for text in graphs?

Yes, but probably not for the reason you might be thinking. Many sans serif typefaces are very geometrical (as opposed to humanist, in which the letters have a modulated stroke as if drawn by a pen tip) and this fits with inherent geometrical nature of most plots.

For example Gotham or Proxima Nova work very well in plots. Helvetica also.

9. Do you recommend a particular 'elements of style' book?

For text layout,

Elements of Typographic Style
https://en.wikipedia.org/wiki/The_Ele...

and for graphics, I recommend you initiall follow the style in the

Points of View

http://mkweb.bcgsc.ca/pointsofview/

especially this one on Elements of Visual Style

https://www.ncbi.nlm.nih.gov/pubmed/2...