If your photos aren’t good enough, then you’re not close enough
— Robert Capa
Papillary thyroid carcinoma (PTC) cells, even though malignant, are still genetically programmed to try to be thyroid follicles and may retain their follicular growth pattern, which appear as circles on cross section. Two diagnostic features of papillary thyroid carcinoma are nuclear clearing and intranuclear cytoplasmic inclusions. The black-and-white image is an artistic treatment of a PTC microscopy image (40×) from one of the Personalized Oncogenomics Program study participants at the BC Cancer Research Center. Superimposed is a Circos plot of 17 genomic fusions involving 17 chromosomes identified in the sample by whole-genome sequencing. Showing through the Circos plot is an enhanced color version of the microscopy image. The original image is from Application of genomics to identify therapeutic targets in recurrent pediatric papillary thyroid carcinoma by Ronsley et al. in the April 2018 issue.
...this special issue provide[s] a glimpse into current cancer precision medicine efforts, reflecting only a microcosm of ... genomics in this bustling space of clinical translation.
John C. Carpten & Elaine R. Mardis
The era of precision oncogenomics
Mol. Case Stud. (2018) 4(2).
I've previously created art based on POG data—posters to celebrate the program's 5-year anniversary.
I've previously taken a more fine-art approach to cover design, such for those of Nature, Genome Research and Trends in Genetics. I've used microscopy images to create a cover for PNAS—the one that made biology look like astrophysics—and thought that this is kind of material I'd start with for the MCS cover.
When I look at these kind of images, I have basically no idea what I'm looking at. Sure, I know this is life at tiny scale but I am not a pathologist. This helps me greatly.
Instead, I see color, shapes, and contrast. I hunt for patterns that would make for an interesting visual, without necessarily trying to communicate any of the science behind that—the paper does a much better job at this than I ever could. It's largely a process driven by intuition and my desire to see distinct visual patterns at different length scales with some symmetry, ideally broken in a pleasing way. Vague, I know.
Images of different regions of the same slide, at the same magnification, can have very different levels of visual engagement (for the non-specialist). Just compare the two images below.
The slide on the left really caught my eye. It had the right proportion of tiny, small, medium and large things.
The black-and-white version was obtained by solarizing the image. There are both color and black-and-white options for solarization, a method in which various tones of the image are remapped in brightness.
And here's the first black-and-white take.
This looked good but a bit dark. I handled this by lightening the tone, differently depending on the element in the image. I also wanted to bring out more details in the internal structure of the cells. This was achieved by applying an otherwise aggressive sharpening mask.
I was quite happy with this result. The combination of solarization and sharpening created a large variety of patterns inside the cells. My brain fought hard to see faces in them.
Because I had slides at different magnifications, I created a design in which three slides at 10, 20 and 40 × were composited together so that from left to right the magnification increased across the image. The effect is subtle—you can easily miss it, which is the point.
I had pretty high hopes for these black-and-white versions. Previous covers in MCS have been colorful, though, so I thought to provide a color option.
For the color version, I wanted to give the colors more punch. For sure.
I also wanted to emphasize the details, like for the black-and-white image.
The first process step of the color slide was done using 5 Nik filters, applied in succession: dark contrast, tonal contrast, sunlight, polarization and detail extractor. The effects of the stack of these filters is shown on the original image below. The whole image is shown and in each strip the filters are stacked.
Here's the full image with the 5 Nik filters applied.
Not there yet, though. I added more sharpening (more than I've ever used before, so I felt a little weird, but got over it quickly). The colors were punched up too—I wanted more contrast between the blue and red areas and transform the reds a little into oranges.
If it looks like the blue areas are popping out of the image, that's the effect of the emboss filter.
The editors asked me to encorporate a Circos image in the final design. This was tricky—I had spent a lot of time up to now fiddling with extracting patterns and textures from the images.
Something as geometrical and rational as a data graphic would alter the personality of the design. But, the goal of artistic collaboration is always to find a way, so I took some gene fusions that were found in the sample with our structural variant pipeline and created a bare-bones Circos image out of them.
This was then superimposed on the image and emphasized by using the color design inside the circle and black-and-white design outside.
It's always fun to invert images and see what happens.
Two-level factorial experiments, in which all combinations of multiple factor levels are used, efficiently estimate factor effects and detect interactions—desirable statistical qualities that can provide deep insight into a system.
They offer two benefits over the widely used one-factor-at-a-time (OFAT) experiments: efficiency and ability to detect interactions.
Since the number of factor combinations can quickly increase, one approach is to model only some of the factorial effects using empirically-validated assumptions of effect sparsity and effect hierarchy. Effect sparsity tells us that in factorial experiments most of the factorial terms are likely to be unimportant. Effect hierarchy tells us that low-order terms (e.g. main effects) tend to be larger than higher-order terms (e.g. two-factor or three-factor interactions).
Smucker, B., Krzywinski, M. & Altman, N. (2019) Points of significance: Two-level factorial experiments Nature Methods 16:211–212.
Krzywinski, M. & Altman, N. (2014) Points of significance: Designing comparative experiments.. Nature Methods 11:597–598.
Celebrate `\pi` Day (March 14th) and set out on an exploration explore accents unknown (to you)!
This year is purely typographical, with something for everyone. Hundreds of digits and hundreds of languages.
A special kids' edition merges math with color and fat fonts.
One moment you're
:) and the next you're
Make sense of it all with my Tree of Emotional life—a hierarchical account of how we feel.
One of my color tools, the
colorsnap application snaps colors in an image to a set of reference colors and reports their proportion.
Below is Times Square rendered using the colors of the MTA subway lines.
Drugs could be more effective if taken when the genetic proteins they target are most active.
Design tip: rediscover CMYK primaries.
Ruben et al. A database of tissue-specific rhythmically expressed human genes has potential applications in circadian medicine Science Translational Medicine 10 Issue 458, eaat8806.
We focus on the important distinction between confidence intervals, typically used to express uncertainty of a sampling statistic such as the mean and, prediction and tolerance intervals, used to make statements about the next value to be drawn from the population.
Confidence intervals provide coverage of a single point—the population mean—with the assurance that the probability of non-coverage is some acceptable value (e.g. 0.05). On the other hand, prediction and tolerance intervals both give information about typical values from the population and the percentage of the population expected to be in the interval. For example, a tolerance interval can be configured to tell us what fraction of sampled values (e.g. 95%) will fall into an interval some fraction of the time (e.g. 95%).
Altman, N. & Krzywinski, M. (2018) Points of significance: Predicting with confidence and tolerance Nature Methods 15:843–844.
Krzywinski, M. & Altman, N. (2013) Points of significance: Importance of being uncertain. Nature Methods 10:809–810.