The Visual Representation
of the Human Genome
David Stairs
Introduction
On June 26, 2000, before foreign dignitaries and hundreds of dis-
tinguished heads of government, business, and science, J. Craig
Venter and Francis Col ins joined President Bill Clinton to jointly
announce the first sequencing of the human genome. As remem-
bered by Venter, “On the great day, happily, all the rivalries were
swept aside by everyone’s feeling of being part of an historic
achievement… As the President, Francis, and I walked together
from the hall into the East Room of the White House, a band struck
up ‘Hail to the Chief,’ and we entered to face a standing ovation.
Two large plasma screens carried a live video linkup to Downing
Street and the British Prime Minister, Tony Blair.”1 Thus is
described a moment of relative calm in one of modern science’s
most contentious rivalries: the one between the National Center for
Human Genome Research headed by Col ins, and Venter’s com-
pany, Celera Genomics. As history now remembers it, it was a pre-
mature announcement of an only partial y completed sequence.
The government program had been launched in the late
1980s, aiming for a completion date of 2005 for sequencing the
genome. As the 1990s rol ed on, the sprawling public program with
several labs spread across America, Britain, and Japan had made
mapping a prerequisite to sequencing, but it clearly was proceed-
ing too slowly to reach its goal. Meanwhile, Venter’s innovation—a
whole genome shotgun assembly sequencing method—eventual y
proved both its quality and speed on efforts to sequence Haemophi-
lus influenzae in 1995 and Drosophila melanogaster in 2000. Eventu-
al y, the government’s program and Celara col aborated to present
a partial y sequenced genome at the White House.
By 2007, Venter released a complete sequence of his own
genome—an amazingly complex visual mapping of all 46 chromo-
somes. How this map came to be, its predecessors, and its signifi-
cance to the history of both biology and design are the topics of
this paper.
1 Craig Venter, A Life Decoded, (New York:
Penguin, 2007), 311-12.
© 2012 Massachusetts Institute of Technology
57
DesignIssues: Volume 28, Number 4 Autumn 2012
Informatics vs. Infographics
The opening line of Edward Tufte’s classic text, The Visual Display
of Quantitative Information, states that “Excellence in statistical
graphics consists of complex ideas communicated with clarity, pre-
cision, and efficiency.”2 The hal mark of communication design is
its ability to visual y represent data for the purpose of easy com-
prehension. Here, one might think of how dramatically Henry
Beck’s 1931 Underground Map for London Transport changed such
representations, or of the way graphic user interfaces (GUIs) pro-
pel ed the PC revolution in the 1990s. Graphical y presented data is
also important to scientific visualization. These days, most data
visualization is computer generated. Unfortunately, the explosion
Figure 1
A Punnett square demonstrating heritability
of computer graphics has ushered in such a trivialization of visual-
of dominant and recessive characteristics
ized data that a distinction must be made between popular and
after Mendel.
often rhetorical infographics, of the sort one sees in the daily paper,
and informatics, or the scientific visualization of data.
I take as my point of departure for scientific informatics
Dmitri Mendeleev’s periodic table of the elements. During the
1860s, several people were trying to make sense of the relation-
ships between the then-known elements. Mendeleev equated the
chemical properties of elements with their atomic weights. His
visual representation was a simple table, the rows and columns of
which are familiar to anyone who today uses the table function of
a word processing program. Mendeleev’s insightful groupings
enabled him to actual y predict the existence of as yet undiscovered
elements. However, visualizing his concept in graphical form
didn’t require that he have any special skill in art.
Over the years, many variations on Mendeleev’s concept
have been rendered—some in table form, some in other formats.
And these days, of course, we find trivialized variations on the
periodic table concept for everything from desserts to video game
characters to commodities returns.
Another, equal y famous example of scientific informatics
is the Punnett Square (see Figure 1). After 1900, Wil iam Bateson
had Gregor Mendel’s 1865 paper on plant hybridization translated
into English. Working with Bateson, Reginald C. Punnett helped
establish genetics at Cambridge, developing his famous square
diagram as a visual means of predicting the outcome of a cross
between two al eles (i.e., forms of a gene) thereby determining the
probability of the offspring’s genotype. Both the periodic table of
elements and the Punnett-square of pea al eles, known to genera-
tions of school children, are examples of graphic diagrams that
were general y accepted and expanded upon by subsequent gener-
ations of researchers—a form of group consensus that has defined
scientific informatics for over a century.
2 Edward R. Tufte, The Visual Display of
Quantitative Information (Cheshire, CT:
Graphics Press, 1983),13.
58
DesignIssues: Volume 28, Number 4 Autumn 2012
The First Genetic Maps
When I was a 16-year-old biology student at Christian Brothers
Academy in Syracuse, NY, Brother George Mason guided us
through the experimental method Thomas Hunt Morgan and his
assistants had developed at Columbia 60 years earlier. Our subject,
of course, was the Drosophila fruit fly, and our goal was to trace
genetic inheritance through eye color. Curiously, a vision-related
characteristic—color blindness—was the first discovered gene-
linked human disability.
In 1890, Weismann had described the process of meiosis,
or how germ cells recombine. Researchers had known of the
existence of chromosomes in the cell nucleus, but for a number of
years they had been puzzled that sperm and egg cel s were hap-
loid, containing only half the number of necessary chromosomes
for reproduction. Some biologists were horrified by the idea of
Mendel’s theory of heredity, suggesting hidden, or recessive, char-
acteristics rather than qualities based on visible characteristics.
Then, in 1895, Wilson wrote: “…inheritance may, perhaps, be
affected by the physical transmission of a particular chemical com-
pound from parent to offspring.”3 He was referring to DNA, the
Rosetta Stone of genetics. By 1900, genes were just beginning to be
linked to chromosomes, but the characterization of DNA and the
means for observing these sub-microscopic entities were still many
years in the future.
By 1902, after researchers increased their observations of
meiosis, Walter Sutter not only determined the stages of meiosis,
but also made a crucial observation—that chromosomes come in
matched pairs, based on size and shape. He called these pairs
homologous (same proportion) chromosomes. It was later deter-
mined that these “homologs” also carry the same genes but that
these genes, because of mutations, may vary in form (e.g., eye
color). These different forms of a gene were cal ed alleles. Moreover,
it was observed that these homologs pair during meiosis and
exchange genetic material between themselves in a process Morgan
cal ed “crossing over.” Identifying this exchange of genetic material
would be crucial to all future understandings of inheritance.
In 1910, Alfred Sturtevant was an undergraduate student
who volunteered in Morgan’s lab at Columbia, the famous “fly
room.” Mendel’s 1865 essays on the inherited traits of peas had
been rediscovered at the turn of the century by Hugo de Vries
and Carl Correns, and only four years had passed since Bateson
introduced the term “genetics” at the Royal Horticultural Society’s
convention in 1906. At that time, the narrow portion of the chro-
mosome, or centromere, was known and would become an impor-
tant reference point for what fol owed. One day Sturtevant had a
3 Alfred Sturtevant, A History of Genetics
(New York: Harper & Row, 1965),104.
DesignIssues: Volume 28, Number 4 Autumn 2012
59
Figure 2
Linkage Map.
flash of insight. As he described it, “I suddenly realized that the
variations in the strength of linkage already attributed by Morgan
to difference in the spatial separation of the gene offered the possi-
bility of determining sequence in the linear dimensions of a chro-
mosome. I went home and spent most of the night in producing the
first chromosome map.”4
Sturtevant’s insight was of a spatial nature; he had noticed
that the frequency with which two genes recombine relates to their
distance from one another on their chromosome. This insight gen-
erated the need to visualize that distance diagrammatical y. Genes
with greater physical separation are likelier to mix during meiosis,
while those closer together are more likely to be inherited together,
as if “linked.” Greater distance equated to an increased chance of
recombination frequency. Using recombined frequencies, Sturte-
vant drew a linear diagram, thus creating the first “linkage map”
showing relative gene location.
A linkage map unit, called a centimorgan in honor of
Morgan, equals a 1% (1 in 100) recombination frequency, based on
the recombination rate of same-site homologous chromosomes
during meiosis. A modern example of Sturtevant’s mapping tech-
nique appears in Figure 2.
The relative locations of genes and gene markers are first
represented as a small fragment, 20,000 base pairs of the gene
ET319246. This gene is itself one of several genes on the longer
range 1.2 mil ion base pair fragment pul ed from the linkage map.
The visual technique of representing increased orders of resolution
by linear nesting, as il ustrated here, is an example of how Sturte-
vant’s idea was later expanded.
From Sturtevant’s initial insight and mapping, research
continued to develop; according to Sturtevant, “The first major
undertaking after 1913 was the mapping of the new genes as they
became available. Here again, while we all took part, it was
4 Peter J. Russell, Stephen L. Wolfe,
Bridges who did most of the spadework, and who gradual y accu-
Paul E. Hertz, Cecie Starr, Biology,
mulated and organized the data to produce the maps; and with the
the Dynamic Science (San Francisco:
Brooks/Cole, 2008), 259.
60
DesignIssues: Volume 28, Number 4 Autumn 2012
carefully planned multiple mutant stocks with conveniently
located markers, it gradual y became possible to work with a preci-
Figure 3
sion that was heretofore impossible with any other material.”5
Chromosome Map.
These early maps were approximations. They didn’t show
precise distances between genes, but only relative positions.
During the 1920s, Sturtevant did determine that Drosophila’s genes
were in linear order by proving that closely related species had
similar mutations, as a result of crossing over, that were al elic and
therefore probably identical. His proposals of linear arrangement
resulted in the evolution from physical maps through linkage
maps to our classic conception of chromosome maps, or ideograms.
In the chromosome map shown in Figure 3, a much larger
distance is represented because chromosomes contain many genes.
The centromere is pictured at the middle of Chromosome 1. Five of
the approximately 4,000 genes that have been mapped on this chro-
mosome are shown in their relative locations. Linkage suggests
that, during the crossing over that occurs at meiosis, the genes at
greater distance from the center are likelier to exchange material.
Molecular Biology, and Beyond
In the years before and after World War II, scientific attention was
turning from high-energy physics, which had been the darling
discipline of science for decades, to biology. Max Delbruck, a
German physicist, moved to California in 1937 to pursue his interest
in biology at Caltech. He began to study the viruses of bacteria, or
“phages,” and was an early proponent of applying mathematics to
make quantitative predictions in biological experiments. His course
in bacteriophage genetics at Cold Spring Harbor and his promotion,
with Salvador Luria, of the “Phage Group” served as an inspiration
to many young scientists in the early days of molecular biology.
Two of these newcomers, James Watson and Francis Crick,
met at the Cavendish Laboratory of Cambridge University. Their
search for the physical structure of DNA, so colorful y described in
Watson’s 1968 book, The Double Helix, actually relates events that
5 Sturtevant,
A History of Genetics, 53-54.
DesignIssues: Volume 28, Number 4 Autumn 2012
61
took place 15 years earlier. In 1952, they embarked on their now
legendary undertaking. Basing their suppositions on earlier
research by Maurice Wilkins, Watson and Crick struggled to under-
stand the “tetranucleotide.” Competing with a team led by Linus
Pauling, and aided by the brilliant X-ray diffraction spectroscopy of
Rosalind Franklin, the two young scientists eventually correctly
modeled the molecule’s helical structure.
This work, informing decades of subsequent research, did
Figure 4
not directly advance genomic mapping techniques. Rather, in
Automated Sequencing Array. Each of the 98
unlocking DNA’s structural mysteries, Watson and Crick aided
horizontal lines represents one capillary tube
research leading to the rise of molecular engineering as we know
DNA sequence. The banding at the center is
the result of an attached end sequence. The
it. Crick’s “central dogma”—that DNA translates to RNA, which
four bases, Adenine, Cytosine, Guanine, and
organizes protein synthesis—led to efforts not only to locate
Thymine, are represented by four distinct
human genes, but also to discover for what proteins these genes
hues: green, yellow, red, and blue. (courtesy
coded. The confluence of discoveries leading inexorably toward
of Jiping Wang, Central Michigan University).
whole genome sequencing was made possible by other researchers
building on Watson and Crick’s work. During the 1950s, for exam-
ple, Margaret Oakley Dayhoff pioneered the first computerized
databanks of DNA sequences. Her 1965 book, the Atlas of Protein
Sequence and Structure, contained data on all of the 65 then-known
protein sequences.6
Less than a decade after Watson and Crick published their
landmark 1953 paper on the structure of DNA, two Philadelphia
researchers, Peter Nowell and David Hungerford, noticed that the
blood cel s of patients with chronic myelogenous leukemia (CML)
had an unusual y tiny chromosome. In 1973, Janet D. Rowley con-
cluded the chromosome was the result of a translocation of mate-
rial between chromosomes 9 and 22. At a time when the genetic
underpinnings of disease were unclear, the discovery of this
abnormality—dubbed the Philadelphia Chromosome—marked the
first time a specific genetic defect was linked to a cancer, paving
the way for drugs that targeted the defect and turned this rare leu-
kemia into a manageable disease.
In 1970, Hamilton Smith purified the first restriction
enzymes. These “molecular scissors” were used to cut DNA at
specific sites. Herbert Boyer and Paul Berg used these enzymes to
splice viral and bacterial DNA in 1972, creating the first recombi-
nant molecule. Boyer’s company, Genentech, founded in 1976,
went on to synthetically manufacture human growth hormone
and insulin.
Frederick Sanger in the United Kingdom and Walter Gilbert
in the United States independently developed new techniques for
sequencing DNA using gel electrophoresis—a technique later
superseded by automated sequencing (see Figure 4). At this point,
6 Molly Fitzgerald-Hayes and Frieda
Reichsman, DNA and Biotechnology
(Burlington, MA: Academic Press,
2009), 150.
62
DesignIssues: Volume 28, Number 4 Autumn 2012
Figure 5
A chromatogram, or trace file, is what
researchers actually look at when reading a
sequence array. Amplitudes of each of the
four base pairs—A, T, C, and G— are repre-
sented by a different color. (courtesy of Eric
Linton, Central Michigan University).
peeking behind the veil of nature to read the specific order of the
amino acids that code human proteins final y became possible. In
1977, PhiX174 became the first organism to be ful y sequenced. Stil
largely a matter of guessing, identifying the location of genes by
custodial “tagging”—first with radiological substances and then
with fluorescent dye tags—eventual y led to automatic sequence
reading (see Figure 5). In these early “BI” (before Internet) days,
gene sequencing was usual y accompanied by experimental char-
acterization of the gene’s biochemical function; so few genes had
been sequenced, an effort was made to understand how they
worked. By 1980, when the U.S. Supreme Court overturned a lower
court ruling al owing genetical y modified organisms to be pat-
ented, the age of molecular engineering had truly arrived.
Computational Genomics
In the mid-1980s, the Department of Energy (DOE) was actively
involved in genomics. Responsible for management of the nation’s
high-energy labs at Los Alamos and Lawrence Livermore, DOE
scientists were interested in radiation’s effect on genes. Breaking
up of DNA into sequenced tag sites (STSs) (i.e., long sequences of
tagged DNA) and expressed sequence tags (ESTs) (i.e., short DNA
copies made from the ends of messenger RNA (mRNA)) had begun
to lead to automated sequencing using the raw computational
power of algorithms that were able to analyze large-volume
throughput of sequence data. DOE scientists and engineers—mod-
elers of the chaos of nuclear explosions—oversaw some of the
nation’s most formidable computing power, and today they stil
run the Joint Genome Institute (www.jgi.doe.gov/).
GenBank, which began in 1982 as a Los Alamos project
and is now housed at the National Center for Biotechnology
Information (www.ncbi.nlm.nih.gov/), was moved to the National
Institutes of Health in Bethesda, MD, at the beginning of the gov-
ernment-sponsored Human Genome Project. James Watson, who
joined as the project’s first director from 1988 until 1992, had by that
time headed up the research institute at Cold Spring Harbor for 20
years, and research there had uncovered the genetic roots of many
human diseases, including cancer. GenBank was designed as a
DesignIssues: Volume 28, Number 4 Autumn 2012
63
Figure 6
Linear representation of a bacterial gene
measured in 20,000 base pair increments.
Genes are color-coded. Top half runs 5’-3’,
bottom half is reversed.
repository for any published genomic sequences, but because the
rate at which they were published was limited by techniques then
available, the archive was small and intensely studied.
Much of the work at that point was theoretical because of
the paucity of data, says Owen Smith, of the University of Mary-
land, who worked with Craig Venter in the early 1990s. “There
were people asking genomic questions at the bench…like trying to
figure out the amount of DNA content in a cell, the number of
chromosomes, the rate of mutation, and thinking about what a
genome was with a chalkboard.”7 People were thinking about
DNA in electrical terms, like a signal in a wire, and researching
various physical means for measuring its content. “There was a
point in history,” he says, in which “basical y every sequence in
the public archive had real meaning, as in it was intensively stud-
ied with bench experiments. Now all that data is an endangered
species because it is swamped out by the high throughput data
coming from genome projects.”8
This perspective is corroborated by Gary Zweiger, who in
Transducing the Genome writes:
In the early days of molecular genetics, genes were
identified on the basis of their function, but when
sequences began gushing into the databases like water
from a hydrant, designations of function began to lag.
Currently, there are many thousands of these orphaned
genes, poor un-christened protein-coders that are
nonetheless rich in concealed information.9
In those “early days,” a variety of approaches were used to visually
represent genomes. As had been the case for decades, some
researchers, following Sturtevant, used the linear model (see Figure
6). This model was not a matter of chance. DNA has a natural direc-
tional linearity, replicating with what scientists call a 5’ (5-prime) to
3’ (3-prime) polarity, or from the phosphate end to the hydroxide
end. According to Eric Linton, a biologist at Central Michigan Uni-
versity, the convention of linear representation arises out of a con-
sensus among scientists for purposes of comparative research. A
graphic that follows the inherent logic of the substance represented
is more likely to be universally accepted and gradually improved.
Graphic notation of genetic information also can often be
7 Personal correspondence (March 4,
found in a circular format. Because bacterial chromosomes natu-
2011).
8 Ibid.
ral y assume a circle, representing their genomes in circular form
9 Gary Zweiger, Transducing the Genome
(New York: McGraw-Hill, 2001), 140.
64
DesignIssues: Volume 28, Number 4 Autumn 2012
Figure 7
Chloroplast genome (courtesy of Eric Linton,
Central Michigan University).
is logical. In Figure 7, for instance, a double-stranded representa-
tion of the chloroplast genome shows the positive 5’ strand on the
outside winding clockwise and the negative 3’ strand on the inside
winding counter-clockwise. In both cases, all the currently known
genes are represented.
Bioinformatics: The New Means for Representing
the Human Genome
On February 16, 2001, nearly eight months after the historic
announcement at the White House, a scientific paper with 274 co-
authors, titled “The Sequence of the Human Genome,” finally
appeared in Science. A companion article representing the
research findings of the public program had appeared the previ-
ous day in the British journal, Nature. Closely reasoned, gener-
ously footnoted, and profusely il ustrated with tables and figures,
the Science paper, with Craig Venter listed as lead author, was an
immediate sensation. Citing 134 preceding articles, Venter’s Celera
Science article has itself been cited 5,056 times and, like Watson
and Crick’s article before it, has gone on to immortality in the
annals of scientific literature.
Sequencing the genome in the lightning-fast time of approx-
imately nine months, Venter’s whole genome shotgun assembly
sequencing, a method that essential y shredded and reassembled
the DNA of five voluntary individuals using banks of powerful
automatic sequencing devices, was an attempt at the then-incon-
ceivable task of creating a high-resolution map that not only
located gene-dense regions on human chromosomes, but also
described gaps in the sequencing process.
DesignIssues: Volume 28, Number 4 Autumn 2012
65
Figure 8
Although the process of transcribing human genes and
Micro-detail of Chromosome 1 on the Venter
expressing proteins is fiendishly elaborate, researchers were sur-
map, complete with nested, labeled known
prised to find that it is quite a bit simpler than they had imagined.
genes. Forward or + polarity phosphate is
5’-3’; reverse is 3’-5’. The rule at top is in
The number of human genes is now known to hover around
million, or mega base pairs. (reproduced by
25,000, although scientists once thought they might find as many
permission, J. Craig Venter Institute, under a
as 150,000; these genes also are governed by fewer mutations than
Creative Commons license from PLoS Biology)
once thought possible. These variations, known as single-nucleo-
tide polymorphisms (SNPs), effect changes in gene expression
among people. Code readers discovered that “less than 1% of SNPs
affect protein function, resulting in an estimate that only thou-
sands, not millions, of genetic variations may contribute to the
structural diversity of human proteins.”10
The ultimate purpose of the research, of course, is not
merely to delineate a particular method. The location and enumer-
ation of human genes is crucial to understanding how genes func-
tion and how the genome evolves. In addition, qualitative
observations, including the fact that proteins with disease associa-
tions belong to duplicated segments of genetic code, have
advanced our understanding of the role that genes play in human
il ness. The 2007 “Diploid Genome Sequence of J. Craig Venter”
(see Figure 8) resulted in the visual representation of a high-resolu-
tion map for all of Venter’s 46 chromosomes, in both the 5’ and 3’
directions. It utilized what are cal ed scaffolds, or long segments of
nucleotides clearly defined by location. These scaffolds were fur-
ther defined by a process of extremely detailed, linear nested,
information overlap, resulting in a remarkable compression of
10 J. Craig Venter et al., “The Sequence
of the Human Genome,” Science 291
(2001): 1330.
66
DesignIssues: Volume 28, Number 4 Autumn 2012
visual y represented data that was nearly able to visual y approxi-
mate the coded representation of the submicroscopic amino acids
themselves. This information exists in printed form, in a wall
poster 40 inches wide by 60 inches high, and it also is available as a
high-resolution 88MB PDF that al ows for a 10x zoom ratio.
This map visual y represents the entire three-bil ion-base
genome, each chromosome in mega-base pairs with its introns and
exons. The exons are the areas of the genome that contain genetic
coding, and they are shown here with their nested pull-outs of
currently known genes. Venter’s sequence was hailed by Design
Observer on September 10, 2007, as “…one of the most complex
single infographics ever created”—a left-handed compliment at
best, given that it represents the culmination of a century of
genetic informatics.
Reading and effectively interpreting the highly specialized
information in this map takes a trained eye, but even with a rudi-
mentary understanding of the science, viewers can begin to under-
stand it. It reads left to right and top to bottom in a linear order,
reminiscent of Sturtevant’s early physical maps. I’ve cross-checked
it against NCBI’s Map Viewer for Chromosome 1—a fascinating
exercise in comprehending relational complexity.11
As the gaps in the genome have been fil ed, our knowledge
of ourselves has grown exponentially. The partial sequence
released in 2001 has become much more complete, as demonstrated
by Venter’s 2007 visualization, and new discoveries proceed apace.
With this growth in complexity comes new chal enges. Genetic
mapping is so much more detailed than it was a century ago that
Sturtevant (1891-1970), who lived until the dawn of molecular biol-
ogy, would have been amazed. Once an area dominated by gener-
alists, genomics has evolved into an area of high specialization.
With the marriage of life sciences and computation, statisticians
and computer scientists work side by side with cytologists and
geneticists to carry on the work once done by undergrad lab assis-
tants. The efforts of the big pharmaceutical companies to mine
data in pursuit of the next generation of drugs continues; mean-
while, the constantly improving database is online and available
for both researchers and informed amateurs to access and update.
As the study of genetics yields new data, as computational
genomics provides tools for new medical and scientific diagnoses,
and as new methods for treating age-old human maladies are fur-
ther developed, such revelations of nature must be illustrated
by ever deepening, more detailed, and subtle forms of visual infor-
mation. Such bioinformatics are often pioneered by scientists seek-
11 See http://www.ncbi.nlm.nih.gov/
mapview/maps.cgi?TAXID=9606&CHR
ing to represent complex ideas in visual form. Sometimes, as with
=1&MAPS=ugHs,genes,genec[5988.93
Mendeleev’s table, an idea sticks and is adopted by subsequent
%3A8650.78]-r&QSTR=100359407
generations. If such visualizations are, in Edward Tufte’s words,
[gene_id]&QUERY=uid%28-2141675
409%29&CMD=UP (accessed March
15, 2012).
DesignIssues: Volume 28, Number 4 Autumn 2012
67
“…communicated with clarity, precision, and efficiency…,” later
generations of researchers will adopt, revise, and improve them.
The history of the development of genomic graphics is rich—and it
will continue to become richer. As visual creatures, we can learn
effectively from images, graphs, and diagrams, no matter how
complex the data set.
Acknowledgement
The author wishes to especial y thank Dr. Eric Linton of Central
Michigan University for his assistance with this essay.
68
DesignIssues: Volume 28, Number 4 Autumn 2012