Unit chart, one square per person. I chose a unit chart because the question is about how many people in each population carry a genotype, and drawing every one of the 2,504 people as a square shows both the share and the real sample size behind it, which a percentage bar would hide. Each row is one of the 26 populations, labeled on the left and grouped under its continental group in bold. Within a row, squares are sorted from RR to RX to XX, so each row reads like a bar made of people. The x-axis counts people from 0 to about 110, so a shorter row means a smaller sample, not a smaller share. Color encodes the number of X copies on an ordinal scale from light blue (RR, two working copies) to dark navy (XX, no alpha-actinin-3 in fast-twitch muscle). I used one hue that gets darker because the three genotypes are ordered, and the XX genotype is not "good" or "bad": it is linked to less sprint power, and some studies report it more often in endurance athletes. A legend above the chart names the three genotypes, and a right-hand column gives each row's XX percentage, so the exact share can be read without counting squares. Within each continental group, rows are sorted by XX share. The chart shows that XX is essentially absent in all seven African populations (0–1%), common in Europe and East Asia (13–31%), and most common in South Asia (27–40%) and among Peruvians (55%). A unit chart becomes hard to read with many more people per row, and these samples of about 100 people per population leave each percentage uncertain by several points.
Small-multiple parallel coordinates. I chose parallel coordinates because the question compares 26 populations across 18 variables at once, and drawing each population as one line across 18 parallel axes shows its whole genetic profile and makes similar populations bundle together. Each vertical axis is one gene variant, labeled at the bottom with its gene and the trait it is linked to, and the y-axis on every axis is the same: the frequency of the counted allele from 0 to 1. I ordered the axes from the variant that differs most across populations (left) to the one that differs least (right), so the answer reads left to right. Twenty-six overlapping lines in five colors would be unreadable and hard to tell apart for color-blind readers, so I split the chart into five panels, one per continental group. In each panel, that group's populations are drawn in blue and all other populations in light gray for context; a single legend at the top explains the two line styles. The leftmost axes show that the skin-pigment variant (SLC24A5, from 0 to 1.00) and lactase persistence (LCT, from 0 to 0.74) separate populations the most, both peaking in Europeans. The third most different variant is a performance gene, ACTN3 (0.08 in Luhya to 0.75 in Peruvians), while most other performance genes, such as HIF1A, ADRB2 and AMPD1, cluster tightly on the right. The panels also show that African populations have the most distinct profile, for example in BDNF (near 0) and AGT (near 0.9). Lines between axes have no meaning in themselves, and allele frequencies hide which individuals carry which genotypes.
Small-multiple bar charts by number of ε4 copies. Risk from APOE depends mainly on how many copies of the ε4 allele a person carries, so I grouped the six genotypes into three plain categories, no copy, one copy and two copies, and gave each its own panel. I chose horizontal bars because comparing lengths along a common baseline is one of the most accurate ways to read quantities, and long group names fit easily on the left. Each panel has the five continental groups as rows, labeled on the left with their sample sizes and sorted the same way in every panel, from the highest to the lowest share of ε4 carriers. The x-axis in each panel is the share of people in that group, starting at 0%. Each panel has its own scale (up to about 100% for no copy, 50% for one copy and 9% for two copies), which the source line states; a single shared scale would shrink the two-copy bars to slivers, which is the problem the stacked design had. Every bar is labeled with its exact percentage, so readers do not need to estimate from the axis. Color follows the direction of risk: blue for no ε4 copy, red for one copy and dark red for two, with "Lower risk ← ε4 copies → Higher risk" above the panels. The chart shows that 53.9% of people in African populations carry no ε4 copy, compared with 71–84% elsewhere; 38.8% carry one copy, compared with 15–27%; and 7.3% carry two copies, compared with 0.6–2.2%. Because each panel has its own scale, bar lengths can only be compared within a panel, not across panels, and grouping by ε4 copies hides the protective ε2 allele. The chart also does not show that the same ε4 genotype is linked to different levels of Alzheimer's risk in different ancestries.
Together, the three visualizations show that genes linked to exercise and health vary across populations in very different ways. Visualization 1 shows that a single performance variant can be almost absent in one continent and common in another. Visualization 2 puts that in context: the variants that differ most between populations are those shaped by skin pigmentation and diet, and most performance-linked variants differ little, with ACTN3 the main exception. Visualization 3 shows that a health-related variant can follow a different geography again, with the Alzheimer's risk allele most common in African populations. No single chart reveals that each gene has its own pattern, so there is no population that consistently carries the "performance" or "healthy" version across genes. What remains unanswered is what these frequencies mean for individuals: the 1000 Genomes participants have no fitness or health measurements, so these charts show where variants are common, not whether carrying them changes anyone's performance or risk.