Descience

Museum inspired scientific posters

The 2018 North American Congress for Conservation Biology (NACCB) was held in Toronto last week and I presented some of the research I am conducting with WCS Canada and the McCann Lab. I really enjoy designing posters for conferences and as in previous posts I like having a central image that covers the entire poster as the focal image with a narrative on the research and as little text as possible. For this poster however, I needed to show some results but I didn’t want to add figures that would detract from the focal image. I recently had visited the NYC Aquarium and many of the interactive exhibits has panels you could lift to learn more about the topic of interest. Inspired by this design, I decided to make a poster that retains the focal image but by adding panels I could still present figures and make the poster interactive.

In Adobe Illustrator (you can use your favourite design software) I designed the poster to have three panels that you could lift. I left space in the design to fit the figures I wanted to include. I then had the poster printed on heavy paper stock at my local printer. Next, with just a xacto knife and a ruler I made three cuts on the left, right and bottom of where I wanted the panel, leaving the top intact to serve as the hinge.

Once the cuts were made, I printed out the figures I wanted to include using a regular laser jet printer. This was the perhaps the most difficult step as I had to resize the figure a couple of times to ensure they fit inside the panel. When the size of the figure was finalized, I simply glued the figure to the back of the poster.

Now when you lift the panel from the front it exposed the figure underneath. To add context I made sure the text on the flap referred to the figure and added additional information on the figure printout. 

Using a focal image for a poster.

I was honoured to be selected as the graduate student speaker for the Department of Biology at McGill University and rather than write an abstract I decided to make a graphical abstract with some animated images. The graphical abstract is a GIF made following the great tutorial by Table Top Whale. Essentially you are making each frame of the animation as a layer in vector graphic software. You then import the layers into Photoshop and make a frame animation, making one layer visible for each frame. Photoshop will then string the frames together set to whichever frames per second you select. It’s a cool way to add some animation to a static poster.

Visualizations are a great way to display complex data but because it is a static plot you can only show a single set of data points. With Shiny however, you can can create a dynamic plot with an interface that allows you to change your inputs and see the effect on your output live. It’s a great way to explore your entire data set with a single plot. The Shiny website has a series of great tutorials with example data to get you started.

In experiments you often have many factors you are testing against each other which can produce great insight but are associated with data visualization problems. How do you visualize multiple comparisons using only a few dimensions? My colleague and bud KP had a data set (she does not work on Trolls...at least not yet) with multiple comparisons and needed a pair of bar plot panels but also wanted to do comparisons between the panels. Inspired by the data visualizations of FiveThirtyEight, a great statistics blog, I suggested producing a the two bar plots with a shared axis to facilitate comparisons between the two plots.

But now how to make this in R. Essentially, you are producing two horizontal bar plots and then merging them with arrangeGrob(). For one of the plots you also have to reverse the axis with scale_y_reverse(), but the key to making them look like one plot is to have a empty panel.back ground and manipulating the plot margins.

In ggplot2 plot.margin allows you to set the size of the margins of your plot. If we want to reduce the distance between two plots that are arranged in two columns then we have to change the right and left margins in plot.margin(). The syntax is plot.margin = unit (c(with the sizes of the top, right, bottom, and left margins), "insert unit here"). Here we reduced the right and left margins to negatives to have the plots almost overlap. For your own graphs you might have to change these dimensions based on the size of your plots. Alternatively, you could also manipulate the distance between the plots with an image editing software like Illustrator once you produce a raw plot in R. You can find the code for the plot including annotations HERE

I'm always looking for new blog post topics so if you have a data visualization challenge feel free to contact me over email or Twitter.

Stories usually have a linear chronology. Your stories in science should be no exception. I like to use timelines in my presentations because they help me tell the story of my research. I usually make my timelines in Adobe Illustrator, but the advantage of making them in R is that you can scale the components of your timeline to display data. In the example above I made a timeline of the different organisms I have worked with throughout my undergraduate and graduate career. Here I scaled the size of the points to indicate how long I worked on that organism and colour coded them based on a binary vertebrate/invertebrate classification. You can find the full annotated code HERE.

That's a big cat! Bar graphs are popular because you can visually detect differences between groups. In single variable comparisons you can also visualize differences by scaling the size of a representative icon. Using the Phylopic data base and the Flylopic package you can change the height of your icons to represent the relative differences between two or more groups. In this example I use the Animals sample data set in the MASS package in R. The data set contains the brain and body masses of several animals. Here I wanted the represent the relative brain size of a cat and the much larger triceratops.

I started by assigning each brain weight to an object and then calculating the ratio of the two brain weights. The height of the cat is then scaled as a proportion of the 1 unit height of the triceratops. I used the element_blank( ) function to produce a blank canvas and delimited the plot boundaries using the position vector. You can see this example and the annotated code HERE. 

Lions and tigers and bears! If you have data from three disparate populations you have several options on how to differentiate your data in R or graphing software. You can use different colors, shape or facets to separate the data across multiple plots. Rather than labeling your plot with a title, Fylopic by Scott Chamberlain allows you to add a layer to your ggplot in R with a silhouette of your organism. Fylopic uses the the existing Phylopic data base containing thousands of images of organisms and their taxonomic information. I created the plot by first searching Phylopic for images of my three organisms. Once I found an appropriate image I noted the UUID or identification number from the image URL - it's the series of numbers and letters after "image/"

Seated to the right of me at "Honey, I Shrunk the Audience! (3D)" was a seven year old boy. As virtual objects were seemingly propelled from the screen the boy made attempts to capture the objects. Failing from a seated position he stood on the theater seat and continued in vain. The only thing preventing me from joining him in his quest was the knowledge that the glasses I was wearing were fooling me.

Anaglyph images, the red/cyan 3D pictures in books and posters, use the same mechanism as 3D movies. To create the illusion of depth, two images are simultaneously printed. The red and cyan films in 3D glasses filter the image so that each eye will see offsetted images. Your brain merges the two images and the illusion of depth is created. The addition of depth immerses viewers but it can also be used display a third dimension in data visualizations. 

Multivariate ordinations often have more information that can be expressed in two dimensions on two axes. Depending on the data set, the addition of a third axis could significantly increase the amount of variation explained in the ordination. You can capture that third dimension by simulating depth with an anaglyph graph. You can create your own anaglyph for 3D plots using R with the 'anaglyph' package. You can follow my example script here to produce this plot HERE.

The package allows you to specify the different locations of your points on the x, y and z axis along with the general aesthetics of the plot function. Just replace the x, y and z objects with your data coordinates to make your plot with your data.

You can also produce a 3D plot using the 'ordiplot3d' package in R or your favourite graphing software and making some adjustments in image editing software.

First produce two 3D scatterplots one with 'cyan' coloured points and one with 'red' points. Instructions on how to do this in R are detailed HERE.

Once you have the two plots import them into an image editing software - I used Illustrator. I removed the extraneous box around the plots and axis tick marks. Next, overlay the two images. I suggest you put the images on separate layers. You only need one set of axis so remove them from one of the image layers. You should now have one set of axis and two sets of points stacked on top of one another. Finally, select the red points and shift the group to the right. You may have to make some adjustments to the position of the red points depending on your red/cyan glasses.

Our sense of touch can evoke memories. I can still remember the softness of my dog's fur, how it billowed under my hand. Visualizing texture in print and in presentations can conjure memories associated with that texture and add dimensionality to print. Instead of using white backgrounds in posters, slides or even business cards, why not try adding a subtle textured background? You can find many free backgrounds like this tree bark online.

You can choose a texture that reflects your experiment, field site or study species. I created a texture that combines several elements of my research - food webs, crayfish, networks and ciliates.

You can view the texture in the header for the blog.

I created this texture by creating a swatch in Adobe Illustrator. Instructions on how to create your own can be found here: http://blog.redbubble.com/2012/05/how-to-make-a-repeat-pattern-in-illustrator/

Despite years of presenting and stage time delivering a talk at a conference still makes me very nervous. For me though, the anxiety begins far before the talk -  in the presenters room. I like to use non-traditional fonts in my presentations because they can add distinctiveness, design or contribute to a theme. Thousands of free fonts from designers are available. One of my favorite sites for fonts is 1001 Free Fonts. However, for a computer to recognize the fonts they must be installed on the presentation computer. When I load my presentation up I am always concerned the font will be replaced by a default and the arrangement of words on my slides will be altered. Waiting for your slides to appear intact can be a tense series of seconds.

One way of ensuring the font you used on your slides is maintained is to embed the font. This step saves the font in the presentation and a lot of stress. A step by step process for different versions of PowerPoint is available here: http://support.microsoft.com/kb/826832

However, there are a couple of caveats. The font must be a true-type font (.ttf extension) but most free fonts are. Also this only works on PCs. As a Mac user once I complete my presentation I use the lab's or a friend's PC, install the font and embed.

Movement in presentations adds a dynamic element that can capture an audience's attention and facilitate storytelling. You are however not limited to using the defaults in presentation software which can often be non-cooperative in front of a conference audience. In her book Slide:ology, Nancy Duarte describes how you can create movement in slides by using the push transitions.

You start by creating a template of the scene you are creating. The example  above is a short presentation about the sea lamprey. Here the template is the body of the sea lamprey. 

We set the slide transitions to push right to emulate moving along the body of the lamprey. This allows us to create a scene from the collection of slides. You could use this method to follow a map along your sampling sites, or data oscillating in a plot. Using other push transitions can allow you to create a more complex landscape too. The idea is if you were to arrange your slides in order and look at them from above, the collection of slides should form one cohesive image or scene.

Infographics communicate information effectively because they represent complex information visually. Here data is not represented as raw numbers but rather as graphics which can be quickly processed and understood. In an environment like a conference where your poster is quickly scanned or where a slide is projected for only seconds translating information into a format that can be processed quickly by the audience is important. For example the methods section of an oral or poster presentation can contain complex information on different factors, manipulations and trials. Information here can be distilled by representing your experimental design through icons.

To explain the manipulations in my experiment I used icons of my three trophic levels. Treatments that received more algae were denoted by more algae icons, while lower predation treatments were represented by more zooplankton icons. The use of icons allowed me to reduce the amount of text in this section and visual representation of the methods facilitated quicker processing of the information.